WATER POLLUTION CONTROL RESEARCH SERIES • 12020 DQC 03/71
Polymeric Materials for
Treatment and Recovery
of Petrochemical Wastes
ENVIRONMENTAL PROTECTION AGENCY • WATER QUALITY OFFICE
-------
WATER POLLUTION CONTROL RESEARCH SERIIES
The Water Pollution Control Research Series describes
the results and progress in the control and abatement
of pollution in our Nation's waters. They provide a
central source of information on the research, develop-
ment, and demonstration activities in the Water Quality
Office, Environmental Protection Agency, through inhouse
research and grants and contracts with Federal, State,
and local agencies, research institutions and industrial
organizations.
Inquiries pertaining to Water Pollution Control Research
Reports should be directed to the Head, Project Reports
System, Office of Research and Development, Water Quality
Office, Environmental Protection Agency, Room 1108,
Washington, B.C. 20242.
-------
POLYMERIC MATERIALS FOR
TREATMENT AND RECOVERY OF
PETROCHEMICAL WASTES
by
GULF SOUTH RESEARCH INSTITUTE
New Orleans, Louisiana
for the
WATER QUALITY OFFICE
ENVIRONMENTAL PROTECTION AGENCY
Grant No. 12020 DQC
March 1971
For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402 - Price 70 cents
-------
Environmental Protection Agency
Review Notice
This report has been reviewed by the Environmental
Protection Agency and approved for publication.
Approval does not signify that the contents
necessarily reflect the views and policies of the
Environmental Protection Agency, nor does mention
of trade names or commercial products constitute
endorsement or recommendation for use.
-------
ABSTRACT
Reverse osmosis has been used as a unit operation to study the
recovery of products from industrial waste streams. Precursory
examination of several industrial wastes was performed.
The recovery of glycerin from a petrochemical waste stream
containing inorganics and polyglycerins has been studied in detail
with the results applied to the design of an effective process
scale unit. Membranes employed were asymmetric cellulose acetate
butyrate and cellulose acetate. The pilot scale experimental
studies were performed with tubular membrane modules which readily
accommodated the sample plant stream being studied.
Good separation was achieved operating between 600 and 800
psig for best selectivity. The product throughput rate appeared
the limiting consideration and proved sensitive to increased
turbulence and reduced feed viscosities the latter achieved
by dilution.
The pilot unit data were used to design a countercurrent
multistage battery to achieve even closer separations. It is shown
that sufficient glycerin could be recovered to provide an attractive
return on the required investment.
This report was submitted in fulfillment of EPA Grant No.
12020 DQC from the Water Quality Office, Environmental Protection
Agency to the Louisiana Department of Commerce and Industry. The
experimental work was performed at Gulf South Research Institute,
New Orleans, Louisiana.
-------
CONTENTS
Section
I
II
III
IV
VI
VII
VIII
IX
X
Conclusions
Recommendations
Introduction
Experimental Methods
Apparatus
Process Measurements
Analytical Measurements
Glycerin Waste Stream
Objectives
Membrane Screening
Process Characterization
Effect of Pressure
Effect of Feed Dilution
Effect of Feed Flow Rate
Effect of Feed Temperature
Multistage Treatment
Projection of Plant Scale Operation
Economic Analysis
Propylene Glycol Waste Stream
Other Waste Streams
Acknowledgments
Bibliography
Appendix
1
3
5
8
8
12
12
14
15
15
22
22
32
42
42
45
50
57
59
62
64
65
66
-------
FIGURES
Page
1 FLAT CELLS UNIT SCHEMATIC 9
2 TUBULAR PILOT UNIT SCHEMATIC 10
3 TUBULAR MODULE SECTION 11
4 FLUX vs. PRESSURE CHARACTERISTIC 23
5 WATER ENRICHMENT 29
6 SEPARATION vs. PRESSURE 30
7 DILUTION vs. PERFORMANCE 40
8 FEED FLOW vs. PERFORMANCE 43
9 TEMPERATURE vs. PERFORMANCE 44
10 SINGLE STAGE OPERATION 51
11 MULTISTAGE FORWARD FEED OPERATION 53
12 MULTISTAGE COUNTERCURRENT OPERATION 55
-------
TABLES
No. Page
1 GLYCERIN STREAM COMPOSITION 14
2 PHYSICAL PROPERTIES OF THE GLYCERIN STREAM 15
3 MEMBRANE SCREENING FOR GLYCERIN STREAM 17
(Eastman/Amicon Membranes)
4 MEMBRANE SCREENING FOR GLYCERIN STREAM 18
(Universal Membranes)
5 MEMBRANE SCREENING FOR GLYCERIN STREAM 19
(Tubular Membranes)
6 HEAT TREATMENT OF EASTMAN MEMBRANE 20
7 F-P-R CHARACTERISTICS FOR MODULE U-3 24
8 F-P-R CHARACTERISTICS FOR MODULE AS-2 25
9 F-P-R CHARACTERISTICS FOR MODULE U-4 26
10 GLYCERIN REJECTION 27
11 SELECTIVITY INDICES FOR TUBULAR MODULES 31
12 F-P-R DILUTION CHARACTERISTICS - 1 33
13 F-P-R DILUTION CHARACTERISTICS - 2 34
14 F-P-R DILUTION CHARACTERISTICS - 3 35
15 F-P-R DILUTION CHARACTERISTICS - 4 36
16 F-P-R DILUTION CHARACTERISTICS - 5 37
17 F-P-R DILUTION CHARACTERISTICS - 6 38
18 FLUX INCREASE WITH DILUTION 39
19 EFFECTS OF DILUTION ON F-P-R CHARACTERISTICS 41
20 FIRST STAGE RUN (MODULE AS-2) 46
21 SECOND STAGE RUN (MODULE AS-2) 47
-------
TABLES
No- Page
22 FIRST STAGE RUN (MODULE U-4) 48
23 SECOND STAGE RUN (MODULE U-4) 49
24 MULTISTAGE OPERATIONS 52
25 ECONOMIC ESTIMATION 58
26 DOW WASTE STREAM - 1 60
27 DOW WASTE STREAM - II 61
28 DISPERSE ORGANIC DYE SYSTEM 63
29 REPRODUCIBILITY OF ANALYTICAL RESULTS 68
-------
Section I
CONCLUSIONS
!• The high degree of selectivity necessary for the recovery of
many valuable water soluble constituents from concentrated multi-
component petrochemical waste streams can be achieved with
commercially available reverse osmosis membranes.
2. For the glycerol stream investigated, the membrane throughput
(flux) was lower for the undiluted and unfiltered waste streams
than for simulated waste streams containing the same components
singly. This was due to the higher viscosities of the plant waste,
and the limited feed velocity available for the experiments. These
factors both lead to increased concentration polarization at the
membrane boundary. Both concentration polarization and an
indeterminant extent of particulate fouling must be avoided in
order to achieve the best possible efficiency.
3. Dilution of the glycerol plant feed stream reduced the severity
of concentration polarization and led to marked improvements in
production rate. The loss of selectivity as a result of dilution
was comparatively insignificant.
4. An experimental module which contained turbulence promoters
exhibited a more marked increase in product flux with dilution
ratio, than a module without the promoters. The turbulence promo-
ters reduced the boundary layer through mixing.
5. A two-stage separation using diluted feed stock produced pro-
duct water containing less than 0.3% dissolved solutes. With a
proper choice of first and second stage membranes recovery can be
made more efficient.
6. Three modes of operation were examined for using reverse
osmosis units to recover glycerin, and thereby reduce the waste
levels normally sent to the plant waste disposal stream. These
three modes are representative only, and are not necessarily the
.optimum configuration for any plant. They illustrate the recoveries
that can be achieved.
a. A single-stage reverse osmosis unit capable of recovering 27%
of the available glycerin can be designed and amortized over
1.5 years based on the current glycerin value. 92% of the
inorganics and 94% of the polyglycerines available in the feed,
were removed in the product.
b. A multi-stage recovery unit with forward feed of the raffinate
to subsequent reverse osmosis recovery batteries was examined.
A three-battery unit provides a product stream which contains
31% of the available glycerin. The product stream contained no
inorganics and 3.9% of the available polyglycerines. For some
1.
-------
.glycerin purification units the absence of inorganics is
necessary. The amortization period for such a design is
3.75 years.
c. An alternative multi-stage recovery unit with countercurrent
feed was examined. A two-battery unit provides a product
stream which contains 31% of the available glycerin. The
design would yield a product containing no inorganics and 5.2%
of the available polyglycerines. This is amortized over 4.5
years.
These three designs can be scaled up to recover more glycerin and
reduce effluent loadings. Process design calculations may be made
based on the objectives established from the examples given.
7. Increased feed temperatures (up to 38°C) increased flux, but
with a concurrent reduction in selectivity indices. At temperatures
above 38°C, the membrane flux declined.
8. The treatment of glycerin production wastes by reverse osmosis
resulted in secondary glycerin recovery. Thus, the effluent
loadings are reduced, but the need for final treatment of the
raffinates remains.
2.
-------
Section II
RECOMMENDATIONS
This program has confirmed that reverse osmosis membranes do offer
the degree of selectivity necessary for the recovery and removal of
selected petrochemical waste constituents. To utilize such separation
procedures the design engineer must be aware of the interactions between
the stream characteristics, the membrane properties, and the flow para-
meters of the device being used. For separation of solutes in the
wastes the selectivity of the membrane must be balanced against the
membrane throughput (flux). Improvement of the latter is the most
important factor in the design of recovery systems. Because of this
need, it is recommended that the following factors be investigated in
greater detail:
1. Pretreatment of waste streams to reduce microparticulates which
lead to membrane fouling.
2. Practical methods for increasing the degree of turbulence in the
tubular modules to reduce concentration polarization.
3. The interactions of several solutes in an aqueous stream to allow
better predictions on the rejection of solutes by membranes in
such streams.
4. Development of selective membranes which are stable over a wider
pH range than the currently available cellulose esters. These
could be based on cellulose ethers, which are known to be stable
to alkaline hydrolysis.
5. Multi-stage separations designed to utilize more than one type of
membrane so chosen to offer both product recovery and water
recovery.
6. A simulated field test for the separations detailed here are
recommended. The test protocol should be based on a constant
feed stream quality, with precautions for pre-filtration. Experi-
mental designs should evaluate the effects of recycle steps, feed
dilution followed by recycle, and feed dilution followed by
recycle into vacumm distillation units.
7. A number of other petrochemical waste streams having potential
for recovery are known to exist; screening experiments with a
number of membrane types should be undertaken to evaluate the
potential of this process in those areas.
8. The design of various reverse osmosis unit configurations should
be examined further to explore the effect of differing design
objectives on the processes which can be achieved. For example,
the recovery of glycerin of high purity may have different design
requirements, than the reduction of chemical oxygen demand in the
3.
-------
effluent. Simllarily, the removal of polyglycerin from a solution
of glycerin and inorganic salts can be achieved with different
process design than either of the two alternatives mentioned. Thus
the application of reverse osmosis as a tool in industrial waste
disposal requires examination of a number of design objectives and
evaluation in terms of both economic return and cost of pollution
control.
4.
-------
Section III
INTRODUCTION
This is the final report for the Water Quality Office, Environ-
mental Protection Agency Grant No. 12020 DQC to the Louisiana
Department of Commerce & Industry. The research was carried out at
Gulf South Research Institute, New Orleans, Louisiana, from June 1969
through June 1970. This report describes the experimental and
analytical procedures used in the program and the results obtained
therefrom. The objective of this program was to evaluate newly
developed semi-permeable membranes for use in in-plant water purifi-
cation and/or recovery of valuable constituents from petrochemical
waste streams.
The need for this study can best be exemplified by describing
the water quality problems affecting one section of the Mississippi
River which serves as a major water waste collection stream for a
petrochemical complex between Baton Rouge and New Orleans, Louisiana.
The river serves as the outfall collector for a refining capacity
well in excess of 600,000 barrels of oil per day. In addition, a
varied and large chemical industry is based on these refined products,
and adds to the effluent loadings.
Current levels of organic extractables are between 0.5 and 1.0
ppm. These levels are above the recommended maximum (0.25 ppm). The
average C.O.D. in the river is 50 ppm. Of this amount, 15 to 20 ppm
have been attributed to petrochemical sources. With a flow of
400,000 ft^/sec., this represents 100,000 tons/day of petrochemicals.
At an average price of six cents per pound, this amounts to a loss of
$150,000/day in material. The recovery of useful chemicals is there-
fore not only an ecological problem, but has economic incentives as
well.
An equally important incentive is the development of in-plant
water purification for recycle purposes. The estimation (1) that by
the year 2000 this nation may face a 200-400 billion gallons per day
deficit of economically extractable fresh water gives a clearer idea
of the magnitude of the problem. In this connection, it should be
pointed out that in an optimally designed reverse osmosis system, the
recovery of constituents and the production of potable water go hand-
in-hand .
Basically, osmosis is the flow of a solvent from a solution
of higher solvent concentration to a solution of lower solvent con-
5.
-------
centration across a semi-permeable membrane. The driving force for
solvent flow is the osmotic pressure difference which is proportion-
al to the difference in solute concentration between the two solu-
tions. If external pressure in excess of the osmotic pressure
difference is applied, a reversal of solvent flow occurs, hence the
term "reverse osmosis". By these means, water, containing impurities,
can be forced through a suitable membrane, leaving the impurities
behind in concentrated solution.
The dual objective of product recovery and water purification
was pursued by:
i) A membrane screening program to establish the waste streams
and membranes suitable for trial, and
ii) Characterization of selected systems to derive engineering de-
sign criteria. Membrane screening was undertaken to evaluate
the various types of commercially available membranes for their
ability to separate some prototype solutes. These were based
initially on a disperse dye (Cibacete) as representative of
large solutes with limited solubility; on glycerin as repre-
sentative of a water soluble solute of intermediate molecular
weight; and on NaCl as representative of the ionic low molecular
weight solutes. The relative rejection ratios of these solutes,
together with the hydraulic permeability of the membranes were
used as a preliminary characterization.
Following the screening program, a more detailed study of the
separation and throughput characteristics of specific systems was
undertaken. The operating ranges of pressure, feed concentration,
feed flow rate, and temperature were evaluated. These data are
necessary for subsequent engineering design calculations to be made.
Unfortunately, present understanding of the transport proces-
ses through membranes is not adequate to predict the behavior of
multicomponent systems, even when the transport of the components
singly is known. Agrawal and Sourirajan (2) have suggested a method
for the prediction of membrane performance with multicomponent aqueous
feeds based on a generalized diffusion model (3). The method is said
to be applicable only to multicomponent feeds having a common ion.
However, the solutions under study here do not meet this requirement,
so that completely phenomenological data must be gathered, although a
number of experiments were carried out on the single components for
future reference in a parallel study ( 4 ).
Through the cooperation of Shell Chemical Corporation, we have
studied in detail one aqueous waste stream containing glycerin,
inorganic salts, and other organic solutes. The separation sought
6.
-------
would allow separation of the glycerin (product) from the other
organic and the inorganic solutes, in such a manner that it would
be suitable for glycerol manufacturing process, (4). A related
stream from Dow Chemical Company was also studied, but with less
success.
In addition to the two pilot streams, a simulated process
stream containing disperse dye was evaluated during the initial
stages of this project. The dye was found to be very tractable by
this purification method. Both high recoveries of the expensive
dye and good quality water were obtained with prototype streams,
but further study on this stream could not be undertaken for a
lack of participating interest.
7-
-------
Section IV
EXPERIMENTAL METHODS
APPARATUS.
Two types of reverse osmosis circuits were used in this research.
A small cell test circuit using flat sheet membranes was employed for
screening of membranes to establish their separation efficiencies and
product rates. Once membrane types and stream details had been
selected, the studies were scaled up to a larger reverse osmosis cir-
cuit using tubular membranes arranged into packaged modules of four
tubes each.
1. Flat Cell Reverse Osmosis Unit. This unit was used for initial
evaluations of flat membranes at pressures up to 1500 psi. It
consists of a Milton Roy duplex pump with a capacity of 24 gph
at 1500 psig, pumping through a Nupro 10 micron in-line filter.
The test cells are built according to a General Atomics design,
and use a rectangular flow pattern. Between the filter and the
three test cells,a T-joint connects a Greer 10 cu. in. accumu-
lator pressurized with nitrogen. The line is maintained at
constant pressure with a Model 8-91 Mighty Mite back pressure
regulator. A Brooks flowmeter is in the low pressure return
line following the pressure regulator. The effluent returns to
the feed reservoir via a tap water cooled heat exchanger. By
maintaining the temperature of the return fluid it is possible
to maintain the temperature differential across the circuit at
a constant value. Fig. 1 shows the schematic of the flat plate
cell circuit.
2. Tubular Reverse Osmosis Unit. The flow diagram of this unit
is shown in Fig. 2. It was designed and sold by American
Standard Co. and is scaled to be suitable for pilot evaluations.
The distinguishing feature of this unit is the tubular design
of the membrane module. It consists of four porous 1/2" dia-
meter, 27" long fiberglass reinforced epoxy tubes which in
operation are lined with the membranes to be tested. This unit
can be adapted to larger tubular modules from all suppliers and
could possibly be employed with any membrane module unit.
Solution is circulated through the module and recycled to
the feed tank. A Yarway diaphragm pump capable of delivering
51.5 gph at 2000 psi was used. Both input and return flows are
monitored by flowmeters; flow pulsations are damped by two
accumulators, one on the inlet and another on the outlet side
of the module. The membrane permeate (or product) goes into the
-------
FIGURE -
AC
A - Accumulator
AC - Air Cylinder
BP - Back Pressure Regulator
CF - Concentrate Flowmeter
CS - Flat Cell Bank
FR - Feed Reservoir
FT
6
HE
P
VBL
VBY
Filter
Pressure Guage
Heat Exchanger
Pump
Bleed Valve
By-pass Valve
- Needle Valve
FLAT CELLS UNIT SCHEMATIC
-------
TUBULAR PILOT UNIT SCHEMATIC
PS
G
ROM
FF
CF
A - Accumulator
B - Back Pressure Regulator
CF - Concentrate Flowmeter
FF - Feed Flowmeter
FR - Feed Reservoir
G - Pressure Gauge
HE - Heat Exchanger
P - Pump
PS - Pressure Switch
R - Safety Valve
ROM - Membrane Module
Figure - 2
10.
-------
FIGURE-3
Porous Support Tube With
Replaceable Osmotic Membrane
Fiberglass
Reinforced
Epoxy Tube
Osmotic
Membrane
^Concentrated
Solution
TUBULAR MODULE SECTION
-------
collection shroud, and the concentrated feed inside the tubes re-
turns to the feed tank via a back-pressure regulator. The return
is metered and cooled on the low pressure side. Because of the
change in membrane properties with temperature, it is necessary to
control the temperature to ±2°C to produce reliable data.
The American Standard tubular unit can operate at pressures
up to 2000 psi. However, in this study pressures of 1200 psi
and lower were used, for reasons to be shown later. Some module
suppliers do limit their units to lower pressure with 1/2" i.d.
tubes. With linear velocities up to 50 cm/sec turbulent flow
should occur. Addition of turbulence promoters will increase
turbulence by an order of magnitude. Fig. 3 shows the tubular
membrane module made of epoxy reinforced fiberglass tubes.
Membranes are loaded in the module as integral tubes or as formed
units from flat sheets. The unit is adaptable to various sizes
of tubular modules.
Process Measurements. All data were obtained with experiments
lasting between two to four hours to reach a steady state before
taking a sample. The temperature of the feed solution was main-
tained at 25°C (±2°C). The feed rate was maintained at 0.4 gpm
on the flat cells unit and between 0.85 and 1.0 gpm on the tubular
unit.
Flux (or throughput) and %R, the percent rejection, are the
two major parameters of the operations of reverse osmosis.
Flux: This was computed by measuring the volume of a per-
meated product through a known effective membrane area per unit
time and expressed as Gallons/Square Foot/Day (gfd) at a fixed
pressure.
Rejection (%R): A rejection for an individual constituent
is defined as:
_, _ Feed Concentration - Product Concentration ...-..,
Feed Concentration
Analytical Measurements:
Single solute experiments were monitored by classical analytical
techniques such as differential refractometry, conductivity and
absorbence spectres copy. For each solute, a calibration curve was
constructed and reproducibility determined to be at least a tenth of
one percent or better. Specifically the concentrations of glycerin
and polyglycols were followed with differential refractometry,
12.
-------
sodium chloride concentration was monitored by changes in conductivity
and dye content was followed using spectrophotometric techniques.
Multi-component analysis was necessary for the process waste
stream and required a specific approach. The method employed followed
the suggested procedure used by Shell Chemical Company in their
quality control laboratories. Summarizing, the glycerin content was
determined by a titration method specific for glycerin, the water
component was monitored using a classical Karl Fisher titration or
by weight loss under controlled conditions, the inorganics were
determined by weighing the residue upon ashing and the polyglycerines
were Hetermined from the differences in weight. The detailed proce-
dure for the analysis is described in the appendix. The deviation of
the analysis between multiple test of the same sample is estimated to
be:
Glycerin ±0.1%
Water ±0.5%
Inorganics ±0.1%
Polyglycols ±0.8%
13.
-------
Section V
GLYCERIN WASTE STREAM
A waste stream from the glycerin manufacturing plant of Shell
Chemical Company, Norco, Louisiana, was studied for possible
recovery of glycerin. The aqueous stream contains glycerin, various
polyglycerines (molecular weight > 100) and a mixture of inorganics
(presumably sodium chloride and sodium hydroxide). A detailed review
on processes for manufacture of glycerin is made by N. Shreeve (5).
At present this stream is wasted since the inorganics and the orga-
nics cause fouling of heat exchangers,corrosion of the equipment,
foaming in boilers and cooling towers and affect the efficiency of the
stripping operation. It was, therefore, necessary to reduce these
constituents to a minimum in order to obtain a usable recycle stream.
The waste stream composition is listed in Table 1. The two
samples, obtained at two different times from the plant vary in
composition, especially with respect to glycerin content.
TABLE - 1
GLYCERIN
STREAM COMPOSITION
Constituent Sample No. 1 Sample No. 2
% Wt. % Wt.
Glycerin 1.2 2.7
Polyglycerines 14.2 17.3
(Molecular Weight >100)
Sodium Chloride + Sodium Hydroxide 2.7 3.0
Water 81.9 77.0
The variation may be due to the process upsets or changes in opera-
ting conditions. The stream also contained some suspended matter.
The physical properties of the stream are listed in Table 2.
It is believed that the osmotic pressure of the stream is mainly due
to glycerin and the inorganics, whereas, the viscosity is due to the
polyglycerines present at a relatively higher concentration. It was
necessary to adjust the pH of the waste stream to 7 with phosphoric
acid in order to preserve membrane life.
14.
-------
TABLE - 2
PHYSICAL PROPERTIES OF THE GLYCERIN STREAM
Color - Dark Brown
Odor - None
3
Density gms/cm - 1.05
Viscosity - 1.5
(Centipoises)
PH _8-9
Suspended Matter - Removable by 10
micron filter
Objectives:
The specific objectives to be met in formulating a solution
were:
i) To identify one or more commercially available membranes that
will selectively retain inorganics and polyglycerines letting
glycerin and water pass through at an optimum flux.
ii) Having chosen such membrane(s), to completely characterize the
membrane-waste stream system under single stage and multistage
operating conditions.
iii) Based on the data obtained, to project a feasible process
scale recovery system and to make a primary economic evalua-
tion of such a system.
Membrane Screening:
Three types of commercial membranes were screened for the present
application;they were:
i) Asymmetric Cellulose acetate-butyrate membranes
(supplied by Universal Water Corp.)
ii) Loeb type desalination membranes made from secondary cellulose
acetate (supplied by Eastman Organic Chemicals and American
Standard)
15.
-------
iii) Polyelectrolyte Complex Membranes (supplied by Amicon Co.)
The first two types of membranes are representative of the
developments resulting from a large program directed toward recover-
ing potable water from sea water and brackish water. The details
of the manufacture, structure, and mode of action of such membranes
have been reviewed in a number of Department of Interior publica-
tions available through the U.S. Government printing office.
Because of the great interest in the application of such membranes
to desalination and the resulting growing literature, it is not
necessary to review their properties here. The polyelectrolyte
membranes have been described in a number of publications available
through the manufacturer. Since much of the details of manufacture
are proprietary, no review can be provided.
Sodium Chloride, Glycerol, Polyethylene Glycol (Molecular Weight
380-420), Polyethylene Glycol (Molecular Weight 580-620) and Ciba-
cete Dye were used as representative solutes; the salt represented
inorganics, and Polyethylene Glycols represented other organics.
These choices represented severe test solutes. The throughput (flux)
and separation characteristics of these solutes at 10,000 ppm con-
centration, at 600/800 psig applied pressure and 0.4 gpm flow rate
were obtained. Primarily, the flat cells unit was used for this
purpose, however, the tubular unit was also used when testing mem-
brane modules.
Tables 3 and 4 illustrate the data obtained on the flat plate
cell test circuit with asymmetric cellulose derivatives and Amicon
membranes. Table 5 lists the data obtained on tubular module test
circuit. These data can be used to point out some general properties
of asymmetric membranes to those readers not familiar with their
characteristics. Increased rejection of sodium chloride is generally
coupled with reduced flows. The reduced flow is a result of two
factors; first, it has been extremely difficult to prepare membranes
which exhibit high fluxes, and at the same time are "tight"; i.e.,
prevent NaCl from permeating. Secondly, if a membrane is in fact
"tight" to the solute in question, the resulting semipermeability
(i.e. water permeability, but solute impermeability) results in the
creation of an osmotic pressure, which acts in opposition to the
applied hydrodynamic pressure. To separate the two effects, it is
a general practice to measure water permeability (usually referred
to as hydraulic permeability) in the absence of any solutes. The
decreased water flux found subsequent to the addition of solute to
the test solution can then be compared to the theoretically calcula-
table osmotic pressure. If in fact there is a larger decrease than
can be accounted for by the creation of the osmotic pressure, the
decrease is often a result of compression of the membrane or foul-
ing by high molecular weight materials.
16.
-------
TABLE - 3
MEMBRANE SCREENING FOR GLYCERIN STREAM
EASTMAN/AMICON MEMBRANES
Eastman Membrane
Solute Concentration:
Membrane
No.
RO-89
G-2
G-3
G-4
G-5
G-6
G-7
NaCl
Flux
gfd
27.9
3.4
4.6
41.5
47.0
39.8
34.7
10,000 ppm
Reject
%
88.0
99.6
99.6
0.77
29.0
42.4
53.7
Flux
gfd
26.2
3.0
4.9
33.7
35.7
48.3
38.7
Pressure: 800 psi
Glycerin
Reject
%
79.7
96.6
95.0
9.6
32.8
34.3
59.0
Ami con Membrane
P!
Flux
gfd
26.0
6.3
5.4
43.3
68.9
33.6
32.6
Polyol 400
Reject
95.0
96.5
97.4
31.0
61.7
69.0
88.0
UM-05
6.29
Pressure: 70 psi
(Other test conditions same as above)
6.15 11.9 5.64 88.2
Geometry: Flat Cell
Polyol 600
Flux Rej ect
gfd %
35.2
49.5
45.2
38.2
36.8
63.2
76.5
92.0
6.85
91.3
-------
CO
TABLE - 4
MEMBRANE SCREENING FOR GLYCERIN STREAM UNIVERSAL WATER CORPORATION MEMBRANES
Solute Concentration: 10,000 ppm Geometry: Flat Cell
Membrane NaCl Glycerin Polyol 400 Polyol 600
No.
CAB-171-15
CAB-171-25
IV-64-2
IV-64-2
IV-64-2
IV-64-3
IV-64-3
IV-64-3
IV-64-4
IV-64-4
IV-64-4
IV-64-5
IV-64-5
IV-64-5
(a)
(76°)
(81°)
(76°)
(81°)
(76°)
(81°)
(76°)
(81°)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
(a)
(b)
Flux
gfd
5.7
7.4
6.0
7.9
198.4
212.0
58.0
76.6
30.0
44.9
60.0
78.1
55.0
72.2
20.2
23.0
62.8
69.7
50.1
68.8
15.9
22.5
53.0
63.1
54.2
72.2
18.8
26.2
Reject
%
93.7
98.3
96.0
97.2
1.2
3.6
67.0
69.0
91.0
93.0
13.8
18.7
67.0
73.9
83.2
89.5
9.6
16.7
76.4
77.0
90.6
91.1
12.0
15.7
59.3
62.4
90.1
90.8
= Operating Pressure 600
Flux
gfd
6.2
8.4
6.1
8.0
81.5
62.0
64.6
80.3
34.6
47.2
53.1
53.7
54.5
67.6
22.4
29.7
42.6
47.7
51.1
64.4
17.6
25.3
39.8
48.3
51.8
70.1
21.4
27.0
psig
Reject
%
> 99
> 99
98
99
0
5
52
55
87
89
5
8
60
63
85
84
3
10
50
50
85
86
6
6
55
60
83
84
.0
.0
.2
.0
.9
.9
.2
.4
.6
.2
.7
.8
.2
.7
.2
.1
.8
.5
.2
.5
.6
.4
.0
.6
.6
.4
.1
.1
Flux
gfd
5.8
8.4
6.7
8.7
61.3
70.1
51.5
66.3
36.6
49.4
45.3
57.1
58.9
69.5
25.6
33.4
46.3
50.1
44.1
51.1
18.3
28.3
32.4
40.1
51.9
61.4
21.0
30.4
(b) =
Reject Flux Rejec
% gfd %
> 99.0
> 99.0
> 99.0
> 99.0
4.8
5.9
83.0
84.0
96.2
94.0
17.5
34.8
80.2
81.8
91.8
92.5
17.8
18.0
82.9
84.6
96.8
98.5
17.4
22.3
85.2
87.9
98.7
> 99.0
Operating
6.3
8.4
6.9
9.2
39.6
49.9
43.6
56.0
35.6
45.2
54.1
68.5
46.3
56.6
20.7
27.5
32.0
39.3
40.9
58.4
19.1
26.8
31.8
42.8
41.8
49.4
21.5
27.0
Pressure
> 99
> 99
> 99
> 99
4.3
4.8
91
98
> 99
> 99
18.1
18.8
92.1
92.7
96.2
97.2
29.4
29.5
87.6
90.8
> 99
> 99
21.4
25.8
90.8
93.7
> 99.0
> 99.0
800 psig
-------
TABLE - 5
MEMBRANE SCREENING FOR GLYCERIN STREAM TUBULAR MEMBRANES
Solute Concentration: 10,000 ppm Feed Flow Rate: 0.9 gpm
Membrane Pressure
No.
Amer. Std.
AS-1 CellAcet
AS- 2
Eastman
G-5
Universal Water
U-l CAB
U-2 CAB
U-3 CAB
psig
600
800
800
800
600
800
800
800
NaCl
Flux
gfd
18.9
22.3
61.2
29.4
24.7
33.0
38.8
46.7
Reject
-------
A second general phenomenon to be observed is that as the pressure
is increased, the water flux increases, but the solute flux does not
increase. Since the rejection ratio is a function of the ratio of so-
lute and solvent fluxes, the rejection ratio increases with pressure.
The absence of solute flux response to pressure is the basis for de-
salination by reverse osmosis, since increased pressure thus leads to
greater productivity-, and purer water. But for separations of multi-
solute streams, it can be a disadvantage, as will be shown later. In
the case of mixed solutes, it is true that increased pressure leads to
higher fluxes, and also to higher rejection ratios, but it lowers sepa-
ration efficiencies since all of the solutes are rejected more
effectively.
The third significant phenomenon illustrated in these tables is
the effect of heat treatment on the thermoplastic membranes. The pro-
cedure for preparing membranes from cellulose esters consists of
quenching a multicomponent gel in an aqueous solution. The membrane at
that stage has very high permeability, and only moderate rejection
properties (See IV-64-2 in Table 4). The membranes then undergo a
structural change by being subjected to an aqueous heat treatment which
results in a marked decrease of permeability, an increase in rejection
ratio, and a moderate stabilization against further flux losses (See
IV-64-2 [81] in Table 4). Since the degree of rejection of the various
solutes is not affected identically, this preparation variable offers a
technique by which separation efficiency can be manipulated. The de-
tails of the heat treatments employed for Eastman membranes listed in
Table 3 are given in Table 6.
TABLE - 6
Membrane No. G-2 G-3 G-4 G-5 G-6 G-7
Shrinkage Temp. °C 83 83 - 83 75 75
Time, Seconds 90 120 - 17 240 300
Selections were made for further screening in the tubular unit
based on the flat plate tests. While the membranes that were to be used
in the tubular tests were available in sheet form, the design of the
American Standard module allowed them to be loaded. This was accomp-
lished by the use of narrow membrane strips which were glued around a
withdrawable mandrel to form tubular elements. The purpose of these
trials was to establish if the improved boundary layer properties of
the larger unit were reflected in improved performance. Unfortunately,
the vagaries of reproducing the glue line, heat treatment, and
"shadowing" of an indeterminate area of the membranes precluded a
rigorous answer.
Instead of continuing with further trials to load sheet membranes
into a tubular unit, a decision was made to obtain the best tubular
castings available to the project. The results of the flat sheet
trials were used as specifications to membrane suppliers to guide their
preparations. From the results of the experiments available, it was
concluded that the samples:
20.
-------
IV-64-2 (76°) G-6
IV-64-3 (76°) G-7
IV-64-4 (76°) UM-05
G-5
merited further investigation.
The IV-64 series showed similar properties, and were all repre-
sentative of mild heat treatment. Consequently, Universal Water
Corp. was asked to supply a 7.01 sq ft. tubular membrane module
with the properties similar to the IV-64 series noted above. Upon
receipt this was designated U-l; its properties are shown in Table
5. It is apparent that the flux and rejection properties are
different from the sheet membranes used for specification. The
tubular unit has approximately 50% lower flux, but a compensatory
rise in rejection. Since this increased rejection was found to be
true for all solutes tested individually, there was a loss of
separation. This consideration overrode all other factors, and the
unit was returned for another casting. As a result Universal Water
Corporation resubmitted two units, designated U-2 and U-3, whose
properties are also shown in Table 5. While the separation factors
were still not as good as the sheet submissions, it was decided
that the trials should proceed with U-3 to characterize the actual
plant streams. During initial testing with the waste stream membrane
failure occurred in module U-3. Module U-4 (a replacement for U-3)
was, therefore, procured from Universal Water Corporation. U-4 was
tested at pressures not exceeding 800 psig.
Two submissions from American Standard comprising 1.04 sq. ft.
tubular units were also chosen for evaluation on the plant streams.
These units (AS-1 and AS-2) were cast directly on to the tubular
support; in addition, they contained turbulence promoters, in the
form of 7/16" diameter PVC balls. Table 5 indicates that AS-2 had a
higher flux and a better separation potential than AS-1, although
the latter is more highly rejecting.
The literature provided by the manufacturer of Dialfo membranes
indicated that a good separation of glycerin could be obtained from
the higher polyols. The salt rejection would, however, be essen-
tially zero, so that the glycerin and salt would permeate together
with the water. For the specific purpose contemplated, i.e., refeed
to the glycerin distillation unit, this separation was not advan-
tageous. The unfavorable separation, coupled with the limited range
of pressures over which they could be used, suggested that further
work on these membranes should be deferred.
As a result, the screening experiments yielded two modules
which were then used for process characterization. These were the
AS-2 and U-4 units.
21.
-------
Process Characterization:
The membranes selected in the screening were subjected to a detailed
study of their performance with the glycerin waste stream as a feed.
The study involved effect of change of:
i) Pressure (range: 500 psig - 1200 psig),
ii) Feed concentration (range: 0 to 50% dilution),
iii) Feed velocity (range: 0.5 to 1 gpm), and
iv) Temperature (range: 20°C to 70°C)
on the flux and rejection characteristics of the membrane. This was
termed as "process characterization" because of its potential impor-
tance in the process design of reverse osmosis systems.
Effect of Pressure on the Membrane Performance;
Tables 7-9 inclusive list the Flux-Pressure-Rejection (FPR)
characteristics of modules U-3, AS-2 and U-4 respectively. While the
specified maximum operating pressure for AS-2 was listed as 1500 psi
we limited the trials to 1200 psi in view of the unfortunate experi-
ence with the U-3 module.
Study of the three tables reveals that some general phenomena can
be deduced. The water flux increases almost linearly with pressure;
the rejection of organics and inorganics increase almost linearly with
pressure; the flux decreases markedly from the flux observed with single
component experiments; and the absolute values of the rejections are
altered significantly in the mixture from comparable values tested
singly.
These generalizations can best be illustrated graphically. In
Figure 4 are shown the flux values for units AS-2, U-3 and U-4 as a
function of pressure. Three observations are significant in this
graph. As mentioned earlier there is a linear relationship between
the flux and the pressure, so that an algebraic function can be de-
rived readily; the slope of the line is a measure of the hydraulic
permeability of the membranes forming the module; from the data one
can deduce that the AS-2 membrane is the "tighter" of the three.
The slope of the line is compatible with the description of
flux given by
J = L (AP - ATT) (1)
w p
where Jw is the flux, L the hydraulic permeability, P the applied
pressure gradient, and ATT is related to the concentrations and re-
flection coefficients of the membrane by
22.
-------
OJ
10
8
0
I I
U-4
U-3
I —
0
0 200 400 600 800 1000 1200
Pressure, PSIG
FLUX vs PRESSURE CHARACTERISTIC
FIGURE-4
-------
No.
1
Pressure
psig
600
Flux
gfd
3.58
800
5.2
1000
6.45
1200
8.4
TABLE - 7
FLUX-PRESSURE-REJECTION CHARACTERISTICS
(Original Stream)
MODULE NO. : U-3
MEMBRANE SOURCE : UNIVERSAL WATER CORPORATION
FEED : STREAM NO. 8
FLOW RATE : 0.9 gpm
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Feed
Composition
% w
1.19
14.22
2.66
81.93
1.03
15.33
3.10
80.54
1.13
13.11
3.76
82.00
1.16
13.72
3.57
81.55
Product
Composition
% w
1.0
3.28
0.32
95.48
0.80
2.41
0.15
96.64
0.70
1.08
0.33
97.9
.63
0.73
0.23
98.44
Rejection
%
16.1
76.8
88.0
22.3
84.3
95.2
38.1
91.8
91.2
45.
94.
93.6
-------
TABLE - 8
No.
Pressure
psig
800
Flux
gfd
1.43
1000
2.49
1200
3.21
FLUX-PRESSURE-REJECTION CHARACTERISTICS
(Original Stream)
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
AS-2
AMERICAN STANDARD
STREAM NO. 8 (ORIGINAL)
0.9 gpm
Feed
Composition
% w
1.57
13.02
2.79
82.62
1.54
12.69
2.84
82.63
1.57
13.07
2.89
82.47
Product
Composition
% w
0.69
1.35
0.35
97.61
0.55
0.67
0.27
98.51
0.45
0.53
0.18
98.84
Rejection
%
56.1
89.6
87.5
64.3
94.7
90.5
71.3
95.9
93.8
-------
TABLE - 9
FLUX-PRESSURE-REJECTION CHARACTERISTICS
No. Pressure Flux
psig gfd
500
ts)
ON
600
700
800
1.55
1.61
1.98
2.35
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
U-4
UNIVERSAL WATER CORP.
STREAM NO. 8 (ORIGINAL)
1.0 gpm
Feed
Composition
% w
2.70
17.42
2.99
76.89
2.72
17.64
3.04
76.60
2.65
17.71
2.96
76.68
2.72
17.93
3.11
76.24
Product
Composition
% w
2.70
9.93
1.64
85.73
,65
,45
,09
88.81
2.65
6.85
0.99
89.51
.46
.88
.04
.62
Rejection
%
43.0
45.2
2.6
57.8
64.2
61.3
66.6
9.6
61.6
67.0
-------
ATT « Eo. ' C. (2)
The net driving force is a function of the feed concentration,
applied pressure, and the rejection coefficients for the solutes being
separated. Referring again to Figure 4, the slope of AS-2 is lower
than the corresponding slopes for U-3 and U-4 indicating that the
former has lower intrinsic permeability. The lower permeability is
often accompanied by a "tighter" structure, as indicated by the higher
rejection ratios found with single solute tests (See Table 5). It is
also confirmed by the effective osmotic pressure found by setting Jw
equal to zero in equation (1). For U-3 and U-4 the back osmotic
pressure at the point that Jw = 0 is found to be 160 psi; for AS-2 the
same intercept is found at 475 psi.
From equation (2) it is apparent that it is the sum of the tj-j/Aci
products which contribute to the back pressure. Since the several
solutes are present in varying molar concentrations, and exhibit dif-
ferent reflection coefficients (o^) , it is not an easy task to assign
the relative sources of osmotic back pressure to the solutes. The situa-
tion is further complicated by the non-ideality of the system; i.e., in
eqn (2) above it is assumed that no interaction exists between the solutes
resulting in zero values for the cross coefficients. The mechanical
formation of boundary layers whose concentrations differ significantly
from the bulk concentrations, also confounds the assignment of the osmo-
tic pressure to the individual solutes. Consequently, one can only use
very general guides at this point in directing membrane development work
toward improved separations until such time as the interactions of the
various physical chemical parameters can be isolated.
A direct comparison of some of the factors itemized above is
possible in the case of glycerin separations. By retabulating the re-
jection of this solute from the data when it is a single solute and
when it is present in a mixture of solutes, the comparisons can be
drawn:
TABLE - 10
GLYCERIN REJECTION (%) AT 800 psi
U-3 AS-2
Single Mixture Single Mixture
60.1 22.2 75.5 55.9
The characterization of membrane performance for separation of
multi-component streams requires that several indices must be develop-
ed. The efficacy of the separation can be examined from the point of
alternative which might be accomplished; this is the most pragmatic
way of describing performance, and we shall utilize it in the sections
27.
-------
to follow. For example, the product stream resulting from U-3 is
richer in water than the feed; the extent of this enrichment as a
function of pressure is shown in Figure 5, but the increase from 82 to
98% is not a very sensitive index.
Figure 6 shows the relationship between recovery/rejection and
applied pressure. It is seen that polyglycerines and inorganics re-
jection increased with the pressure thus providing a purer product,
but the higher pressure was not advantageous for the glycerin
recovery since the glycerin recovered in the permeate decreased from
83.9% at 600 psig to 54.4% at 1200 psig.
The three curves in Figure 6 thus summarize the potential of sepa-
ration; both inorganic and organic levels in the permeate can be re-
duced substantially, while only a relatively minor percentage of the
glycerin level is reduced. The permeate thus may be suitable for
reprocessing in the original recovery unit since the deleterious
components - inorganics and higher molecular weight polyols - have
been reduced substantially.
An algebraic index of this type can also be developed in the
following manner if a selectivity index for organics is defined:
S = (% Rejection of Organics - % Rejection of Glycerin)
and an analogous one for inorganics:
S. = (% Rejection of Inorganics - % Rejection of Glycerin)
These parameters are indicative of the best separation of glycerin
attainable. The separation factor does not, however, take into account
productivity, so that this variable must be designated separately.
The data presented thus far on modules U-3 and AS-2 were also
evaluated on the replacement module U-4. This unit exhibited a lower
rejection for all of the solutes than U-3, but unfortunately did not
compensate as well by increased flux. The effective osmotic pressure
found by extrapolation of the Jw value to zero in Figure 6 is again
found to be 160 psi. While these data would appear to be contradic-
tory, i.e., lower flux and low rejection, they are in fact a conse-
quence of the hydrodynamics of the process. Examination of the feed
concentrations for the evaluation runs in Tables 6 and 8 reveals that
the plant effluent under test differed in composition between the two
trials. The U-4 module was in fact tested on a stream containing
higher dissolved total solutes, vis. 23% compared to 18% for U-3. The
determination that the same effective osmotic back pressure was opera-
ting with a much more concentrated feed stream indicates that the re-
flection coefficients must be lower for U-4 than for U-3. That this
increased permeability does not result in increased flux will be shown
to be the result of a marked increase in viscosity of the feed stream.
28.
-------
FIGURE - 5
Module : U-3
Feed Rate : 0.9 GPM
^^ : % Water Enrichmen
100
90
80
70
60
50
40
30
20
10
0
-
"™
—
-
—
—
—
—
—
m
m
m
m
600 800 1000 1200
Pressure , PSIG
WATER ENRICHMENT
29.
-------
MODULE U-3
FLOW RATE 0-9 GPM.
100
^ 80<
LJ
40
20
A GLYCERIN
0 POLYGLYCERINE
a INORGANICS
100
40
TJ
3J
O
20 q
600 800 1000 1200
PRESSURE ,PSIG.
FIGURE -6
SEPARATION vs PRESSURE
30.
-------
The significant factor found with U-4 was that a very concentrated
multicomponent stream could indeed be separated by this technique, and
with quite good efficiency. Using the index developed earlier, (So and
S-jJ , the comparison of modules U-3, AS-2 and U-4 are shown in Table 11.
It is apparent that the U-4 module is as effective in achieving separa-
tions as the original U-3, and that the loss of flux is related more
to the hydrodynamics than to the membrane properties, per se.
TABLE - 11
SELECTIVITY INDICES FOR THE TUBULAR MODULES
Module Pressure Flux Selectivity
No. psig gfd S0 Si
U-3 800 5.20 62.1 73.0
AS-2 1000 2.47 30.2 26.1
U-4 700 1.98 61.4 66.6
A review of the tables presented thus far also indicates that the
most effective separations are carried out at the lower rather than
the higher operating pressures. Consequently, further studies con-
centrated on the pressures of 600-800 psi.
In examining the flux values for the several modules it is quite
apparent that there was a significant decrease in going from a single
solute at 10,000 ppm in water to the plant effluent. Part of the de-
crease at any given pressure is attributable to the back osmotic
pressure, which has already been identified as varying from 160 to
450 psi. However, even if the pressures are corrected to reflect the
osmotic forces there is still a reduction in flux compared to the
single solute tests. Two reasons for this flux decline have been
identified, and the steps necessary to combat the phenomenon can be
outlined.
The crude plant stream was used without any filtration for the
majority of the data reported. On microscopical examination during
the program it was found that particulate matter was present in the
stream, and can form a surface layer which can affect the flux rejec-
tion performance of the system. Some of this foreign material may
arise from precipitation during the neutralization of the feed stream.
The result of such fouling is that the effective area available for
separation is reduced, and a parallel layer of a dynamically formed
membrane is created. Prefiltration of the process stream through a
25 micron mesh filter was found to reduce this contamination signifi-
cantly, but the fouling of the membrane had already been accomplished.
A second cause of flux reduction results from the increased vis-
cosity of the feed stream due to the high solute concentrations. The
31.
-------
increase in viscosity reduced the turbulence and mixing at the solution-
membrane interface. As the solvent permeates the membrane it leaves be-
hind a layer of concentrated solution. If the turbulent mixing is inade-
quate to reduce the concentration in this layer to that of the bulk so-
lution, the flux declines because the concentration gradient determining
the osmotic back pressure (eqn 2) is that gradient which exists at each
membrane interface. If the gradient increases, the osmotic back pressure
increases, and the membrane "sees" a lower applied pressure. This
phenomenon is termed concentration polarization, and is one of the most
significant causes of flux decline in poorly designed systems. From the
description given above it is evident that the magnitude of the concen-
tration polarization is giverned by the geometry of flow, the velocity
of flow, and by the membrane transport characteristics.
Effect of Feed Dilution on the Membrane Performance:
The concentration polarization described above can be reduced
either by an increase in feed flow velocity, or by reduction of the
feed stream viscosity. The temperature limitations for the membrane
impose an upper limit to the lowering of solution viscosity which could
be achieved through higher temperature operation, but dilution of the
feed stream can effect significant lowering of the feed viscosity, and
this avenue was explored in this project. The effect of diluting the
effluent stream is exemplified by the data gathered with the U-4 module,
and tabulated in Tables 12 through 18. The data are summarized in terms
of the effect of dilution on the flux and on the selectivity index
(SQ and S±) in Table 19.
Another way of increasing mixing was by the incorporation of turbu-
lence promoters into the tubes themselves. The AS-2 unit contained
7/16" diameter balls in the 1/2" inside diameter tubes. Localized velo-
cities are increased significantly in this manner. A possible drawback
to the use of such turbulence promoters is the potential that they may
impact on the membrane surface and cause permanent damage. Additionally,
the increased flow resistance creates a pressure drop down the length of
the tube so that power is dissipated, with its attendant cost. The
alternative - increased flow velocity through increased pump capacity -
can be evaluated for each geometry under consideration. The mass transfer
through the membrane and in the bulk solution is also effected by vis-
cosity. It is not possible to separate the boundary layer problem from
this phenomenon. The results indicate major improvements can be achieved
by dilution.
The quantitative idea of the flux increase with dilution for the
Module U-4 may be obtained from Table 18 and Figure 7, and it serves as
an index for the measure of increase in flux. It is observed from Table
18 that a flux increase of more than four-fold is obtained with a 50%
dilution. Appreciable flux increase (almost three-fold) was observed
also at lower levels of dilution (30%). Lower viscosity and lower osmotic
effect as a result of the dilution appear mainly responsible for the
increased flux.
32.
-------
No.
Pressure
psig
400
600
700
•x, 4
800
TABLE - 12
FLUX-PRESSURE-REJECTION (16% DILUTION) CHARACTERISTICS - 1
Flux
gfd
1.71
2.84
3.34
4.22
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
U-4
UNIVERSAL WATER CORP.
STREAM NO. 8
1.0 gpm
Feed
Composition
% w
2.19
13.15
2.26
82.40
2.07
13.50
2.35
82.08
2.14
13.77
2.40
81.69
2.21
15.58
2.63
79.60
Product
Composition
% w
2.10
5.10
0.67
92.13
1.91
3.58
0.41
94.10
1.
2.
.80
,39
0.28
95.53
1.75
3.99
0.32
94.44
Rejection
%
4.1
61.2
60.4
7.7
73.5
82.6
15.9
82.7
88.3
20.8
74.4
87.8
-------
No.
Pressure
psig
500
600
TABLE - 13
FLUX-PRESSURE-REJECTION (25% DILUTION) CHARACTERISTICS - 2
Flux
gfd
3.56
4.25
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
Glycerin
Poly glycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
U-4
UNIVERSAL WATER CORP.
STREAM NO. 8
1.0 gpm
Feed
Composition
% w
1.63
13.56
2.17
82.70
1.87
13.00
2.26
82.88
Product
Composition
% w
1.46
4.04
0.40
94.10
1.48
2.93
0.49
95.10
Rejection
10.4
70.2
81.6
20.9
77.5
78.3
700 5.57 Glycerin
Polyglycerines
Inorganics
Water
1.80
12.99
2.33
82.88
35
31
0.43
95.10
25.0
82.2
81.5
800 6.22 Glycerin
Polyglycerines
Inorganics
Water
1.80
13.24
2.35
82.61
1.
2.
,26
.06
0.37
96.31
30.0
84.4
84.3
-------
TABLE -14
FLUX-PRESSURE-KEJECTION (30% DILUTION) CHARACTERISTICS
- 3
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
U-4
UNIVERSAL WATER CORP.
STREAM NO. 8
1.0 gpm
No.
Pressure
psig
500
OJ
Ln
600
700
800
Flux
gfd
3.86
5.02
6.39
6.65
Glycerin
Polyglycerin.es
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Feed
Composition
% w
1.65
10.53
2.01
85.81
1.61
10.89
2.03
85.47
1.64
13.54
2.11
82.71
1.57
13.94
2.15
82.34
Product
Composition
w
,39
,71
0.46
95.44
1.34
2.11
0.30
96.25
1.
3.
.28
,21
0.27
95.24
1.
2.
.18
.66
0.22
95.94
Rejection
15.8
74.3
77.1
16.8
80.6
85.2
22
76.3
87.2
24.8
80.9
89.8
-------
No.
Pressure
psig
500
600
700
800
TABLE - 15
FLUX-PRESSURE-REJECTION (40% DILUTION) CHARACTERISTICS - 4
Flux
gfd
4.67
6.05
6.99
8.4
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
U-4
UNIVERSAL WATER CORP.
STREAM NO. 8
1.0 gpm
Feed
Composition
% w
1.50
9.96
1.79
86.75
1.52
10.07
1.93
86.48
1.44
10.30
1.96
86.30
1.49
11.24
1.98
86.29
Product
Composition
% w
1.24
2.02
0.33
96.41
1.17
1.63
0.27
96.93
1.12
1.38
0.22
97.28
0.96
1.33
0.15
97.56
Rejection
17.3
79.7
81.6
23
83.8
86.0
22.5
86.6
88.8
35.6
88.2
92.4
-------
TABLE - 16
FLUX-PRESSURE-REJECTION (45% DILUTION) CHARACTERISTICS
MODULE NO. : U-4
MEMBRANE SOURCE : UNIVERSAL WATER CORP.
FEED : STREAM NO. 8
FLOW RATE : 1.0 gpm
- 5
No.
1
Pressure
psig
500
600
700
800
Flux
5.75
6.52
8.1
9.09
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Feed
Composition
% w
1.27
9.77
1.72
87.24
1.09
10.38
1.92
86.61
1.29
10.05
1.74
86.92
1.43
9.96
1.78
86.83
Product
Composition
% w
1.01
2.35
0.28
96.36
0.94
1.68
0.32
97.06
0.83
1.11
0.13
97.93
0.69
1.01
0.10
98.2
Rejection
%
20.5
75.9
83.7
13.9
83.8
83.3
35.7
89.0
92.5
51.7
89.9
94.4
-------
TABLE -17
FLUX-PRESSURE-REJECTION (50% DILUTION) CHARACTERISTICS
- 6
No.
UJ
CO
Pressure
psig
500
600
700
Flux
gfd
5.55
6.72
i.7
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
U-4
UNIVERSAL WATER CORP.
STREAM NO. 8
1.0 gpm
Feed
Composition
% w
1.13
8.84
1.58
88.35
1.08
9.21
1.56
88.15
0.92
9.15
1.57
88.36
Product
Composition
% w
0.82
1.66
0.19
97.33
0.83
1.17
0.11
97.89
0.70
0.98
0.07
98.25
Rejection
27.4
81.2
88.0
23.1
87.3
92.9
23.9
89.3
95.5
-------
TABLE - 18
FLUX INCREASE WITH DILUTION
MODULE NO.
MEMBRANE SOURCE
FEED RATE
PRESSURE
: U-4
: UNIVERSAL WATER CORP.
: 1 gpm
: 600 psig
No.
% Dilution
16
25
30
40
50
Flux
gfd
1.66
2.84
4.25
5.02
6.05
6.72
% Increase
in flux
71
156
202
264
305
39.
-------
Module-U4
Flow Rote - j gpjr»
o
i
X
3
10
50
20 30 40
Dilution- %
Dilution versus Performance
FIGURE -7
40.
-------
TABLE - 19
EFFECT OF DILUTION ON FLUX-PRESSURE-REJECTION RELATIONSHIPS
MODULE NO. : U-4
MEMBRANE SOURCE : UNIVERSAL WATER CORP.
FEED RATE : 1 gpm
Pres- 16% Dilution 25% Dilution 30% Dilution 40% Dilution 45% Dilution 50% Dilution
sure Flux S S. Flux S S. Flux S S. Flux S S. Flux S S. Flux S S.
01 01 01 ox 01 01
400 1.7 56 57 - --_ ___ __ _ __ _ __
500 - - - 3.6 71 60 3.9 61 58 4.7 65 63 5.8 64 56 5.6 61 54
600 2.8 75 66 4.3 57 56 5.0 68 64 6.1 64 61 6.5 70 70 6.7 71 65
700 3.3 74 74 5.6 57 57 6.4 65 54 7.0 66 64 8.1 58 53 8.7 70 65
800 4.2 68 57 6.2 54 54 6.7 65 57 8.4 57 53 9.1 42 38 9.9
-------
The Effect of Feed Flow Rate on Membrane Performance:
The effect of feed flow rate on the performance of module U-4 was
studied. The range of variation of the feed flow rate was 0.5 to 1.0
gpm at 600 psig. The results are plotted in Figure 8 as flux;
S0, and S-£ (defined earlier) vs. feed flow rate.
The flux increased from 4.8 gfd at 0.5 gpm to 7.0 gfd at 0.95 gpm
exponentially. No definite trend was, however, noticed with respect to
rejections and, therefore, S and S-.
The flux increase could be explained on the basis of concentration
polarization effect studied extensively by several writers (8, 9, 10).
The concentration polarization (which is a strong function of feed
concentration and feed viscosity) decreases with higher feed rates due
to better mixing conditions at the membrane-feed solution interface.
This consequently improves the effective driving pressure for the
system and hence the increase in flux. The lack of any such trend with
rejections is hard to explain with the limited data at hand, especially
within the very narrow range of change of feed flow rate studied here.
The feed flow rate was limited to one gallon per minute for the
present experiments due to the limitation of the pilot unit. Parts to
modify the unit, in order to obtain higher feed flows, were ordered
during the grant period. However, these parts were not received in
time to make the necessary modifications and to include higher feed
rates as a part of this study. The data does, however, indicate that
additional increases in flux might be achieved at flow rates exceeding
one gallon per minute.
The Effect of Temperature on Membrane Performance:
The effect of feed temperature on the performance of module U-4 was
studied at a constant feed rate of 0.9 gpm and a pressure of 600 psi.
The feed solution was subjected to temperature increase from 20°C to
70°C. The results are plotted in Figure 9 as Flux, So and S^ versus the
feed temperature.
The flux increased initially with temperature reaching a maximum
of 6.6 gfd at about 38°C and then decreased gradually to 5.7 gfd at 65°C.
The selectivity indices So and S^ decreased steadily with an increase in
temperature. The changes in these membrane properties were found irre-
versible with temperature.
This response is not surprising considering the level of solutes
present in the feed solution. It has been well established that in pure
water, cellulose acetate membranes undergo a glass transition around
75°C. This is a lowering of the same transition that occurs in the dry
membrane by approximately 100°C. The lowering in the transition tempera-
42.
-------
Module-U4
600psig.
8
0.7 0.8 0.9 1.0
Feed Rate-gpm
Feed Flow versus Performance
, FIGURE-8
43.
-------
MODULE-U4
GOOpsig.
£4
O
I
X
Si
Flux
70
60
50
40 £
"o
30
20
10
o
o.
0>
20 30 40 50 60 70
Temperature- °C
Temperature versus Performance
FIGURE-9
44.
-------
ture is attributed to the plasticization that occurs in the presence
of water. Glycerin is a better plasticizer than water and, therefore,
the glass transition in the system under study should occur at a
temperature even lower than that with water. The data shows the glass
transition temperature in the present case to be close to 38°C. The
drop in the flux and selectivities So and S-j^ is due to the change in
the membrane morphology (reduction in the porosity) above the glass
transition temperature. The initial increase in the flux is believed
to be related to the lowered viscosity of the feed with increased
temperature.
Multistage Treatment;
The product from a single stage treatment still contains signifi-
cant amounts of organic and inorganic constituents to be returned to
the process unit. A multistage treatment was considered to overcome
this problem and, therefore, several two-stage treatment studies were
made. The same membrane module was used in the first and the second
stage. The membranes were selected on the basis of maximum separa-
tion for a one pass system and, therefore, may not be the best choice
for the second and subsequent stages. The efficiency of the system,
however, should be adequate to demonstrate the advantages of such a
treatment.
Modules AS-2 and U-4 were subjected to two-stage runs at 1,000 and
600 psig respectively. The flow scheme for these runs involved stepwise
concentration of the diluted feed (Tables 20 and 22) by raffinate
recycle in the first stage, followed by another stepwise concentration
of the combined first stage product to obtain the final product (Tables
21 and 23) in the second stage. Samples were drawn approximately every
three hours in the first stage and approximately one and one-half hours
in the second stage. The run duration was, thus, about twelve hours
for the first stage and about six hours for the second stage.
The Module U-4 exhibited better selectivity over Module AS-2.
The selectivity is mainly dependent on the membrane structures, however,
to an extent, this was also due to the differences in the feed compo-
sitions. The feed to U-4 was lower in total solute concentration.
The turbulence promoter containing Module AS-2, however, offered a
higher flux, notwithstanding the higher rejection characteristics,
when compared with Module U-4. The lower viscosity (resulting mainly
from the lower total solute concentration in the feed) and a higher
level of turbulence (brought about by the turbulence promoters) are
believed to be responsible for this performance.
45.
-------
TABLE - 20
FIRST STAGE RUN
No.
1
Flux
gfd
22.2
19.7
16.8
13.25
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
PRESSURE
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
AS-2
AMERICAN STANDARD
STREAM NO. 8 (INITIAL DILUTION 60%)
0.9 gpm
1000 psig
Feed
Composition
% w
0.56
5.43
1.14
92.87
0.73
6.72
1.45
91.10
0.92
8.54
1.80
88.74
1.04
7.63
2.24
86.09
Product
Composition
% w
0.18
0.16
0.04
99.62
0.25
0.16
0.06
94.53
0.23
0.32
0.08
99.37
0.30
0.51
0.13
99.06
Rejection
67.9
97.1
96.5
65.8
97.6
95.9
75.0
96.3
95.6
71.2
93.3
94.2
-------
TABLE-21
SECOND STAGE RUN
No.
1
Flux
gfd
34.7
34.1
28.4
17.3
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
PRESSURE
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
AS-2
AMERICAN STANDARD
STREAM NO. 8 (COMBINED FIRST STAGE PRODUCT)
0.9 gpm
1000 psig
Feed
Composition
% w
0.20
0.30
0.08
99.42
0.22
0.40
0.09
99.29
0.23
0.57
0.10
99.20
0.28
0.73
0.15
99.89
Product
Composition
% w
0.07
0.02
0.01
97.90
0.09
0.02
0.01
99.89
0.09
0.02
0.01
99.87
0.09
0.10
0.01
99.89
Rejection
65.0
93.3
87.5
59.1
95.1
88.9
60.9
96.5
90.0
67.9
86.3
93.3
-------
TABLE- 22
FIRST STAGE RUN
No.
1
oo
Flux
gfd
8.25
5.91
3.42
1.59
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
PRESSURE
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
Glycerin
Polyglycerines
Inorganics
Water
U-4
UNIVERSAL WATER CORP.
STREAM NO. 8 (INITIAL DILI TION 50%)
1.0 gpm
600 psig
Feed
Composition
% w
1.02
8.30
1.48
89.20
1.13
10.75
2.04
86.08
1.19
11.77
2.37
83.65
1.25
16.29
3.20
79.26
Product
Composition
% w
0.73
1.31
0.19
97.77
0.87
1.67
0.19
97.27
0.98
2.22
0.31
96.49
1.13
4.09
0.51
94.27
Rejection
28.4
84.2
87.2
23.0
84.5
90.7
17.7
81.1
86.9
9.6
74.9
84.1
-------
TABLE - 23
SECOND STAGE RUN
No.
1
MODULE NO.
MEMBRANE SOURCE
FEED
FLOW RATE
PRESSURE
Flux
gfd
19.0 Glycerin
Polyglycerines
Inorganics
Water
16.6 Glycerin
Polyglycerines
Inorganics
Water
16.6 Glycerin
Polyglycerines
Inorganics
Water
15.5 Glycerin
Polyglycerines
Inorganics
Water
U-4
UNIVERSAL WATER CORP.
STREAM NO. 8 (COMBINED FIRST STAGE PRODUCT)
1.0 gpm
600 psig
Rejection
%
32.9
71.4
100.0
37.4
70.5
100.0
27.8
65.8
87.1
27.9
65.8
87.1
Feed
Composition
% w
0.82
2.27
0.16
96.75
0.91
2.81
0.23
0.97
3.33
0.31
95.39
0.97
3.33
0.31
95.39
Product
Composition
% w
0.55
0.65
0
98.8
0.57
0.83
0
0.70
1.14
0.04
98.12
0.70
1.14
0.04
98.12
-------
Projection to Plant Scale Operation:
A projected plant-scale operation can be assembled from the
laboratory experimental results, using the data from the U-4 module,
operated at 600 psig with a feed diluted 40% with water as a base
point. The dilution of the feed, as mentioned in the preceding
section, improves the flux through the membrane but at a cost of
glycerin concentration in the membrane-permeated product (this may
lead to an increased distillation cost in some systems).
Three different models are depicted to illustrate the recovery
of glycerin from the waste stream using both single stage and multi-
stage operation. Various other alternatives are possible depending on
the specification of the final extract composition and characteristics
of the membranes employed, however these three will demonstrate the
applicability of the technique of Reverse Osmosis to the treatment of
glycerin waste streams. The three models are:
i) Single stage operation;
ii) Multistage operation with forward feed;
iii) Multistage countercurrent operation.
The basis of calculations for all three models is based on the
data shown in Tables 20 through 23,and an undiluted feed rate of 10
gallons per minute. The feed composition (percent by weight) of this
undiluted waste stream is shown below:
Water 76%
Polyglycerine 18%
Glycerin 3%
Inorganics 3%
The average values of membrane flux and rejection were estimated from
experimental data using a linear intergration of these parameters as
a function of viscosity and membrane area. Figure 7. illustrates that
such an estimate should yield a conservative value for flux. The mem-
brane area was selected based on a design philosophy to remove the
water of dilution.
Single Stage Operation:
Figure 10 shows the single stage material balance for glycerin
recovery operating at 600 psig and with a 10 gpm (approximately 85
Ibs. per minute) raw feed rate and a 4 gpm (approximately 34 Ibs. per
minute) water added for the reduction of the feed stream viscosity.
50.
-------
W,P,G,I : Ibs./min.
FIGURE-IO
Total
W: <
P'
R ai.7 VooH
W: 64.5 I:
p- is ^ ^
G: 2.6 <
I: 2.6
W:
Feed
)8.5
L5.3
9 A
2.6
34
1
S
•
i
I
i
Ex
W:
P:
I:
Raf
W:
-to p-
G:
I:
tract
33.2
1.0
0.6
0.2
finate
65.3
14.3
2.0
2.4
SINGLE STAGE OPERATION
-------
Under these conditions the average flux through the membrane is esti-
mated to be 5 gallons/ft^/day. Employing 1200 square feet of membrane
area and assuming the rejection ratios to be Glycerin 22%, Polyglycerin
78%, Inorganics 78%, as determined in the experimental test, the water
of dilution can be recovered; in addition to 10% of the raw feed water.
This recovered product contains 27% of the recoverable glycerin along
with a small percentage of the original polyglycerin and inorganics.
For some systems this product stream from a single stage separation
may be adequate for recycle. However, in the particular system under
study the deleterious effects of the inorganics in the glycerin dis-
tillation require their complete removal. This can be accomplished in
a multistage operation.
Multistage Operation with Forward Feed:
In a multistage operation the product from the first stage is
processed through a second stage. The first stage product has a low
total dissolved solids concentration and, therefore, the flux rate in
the second stage is much higher than in the first stage. In this
particular model the raffinates from stages 1 and 2 are combined and
diluted with water to obtain a solution viscosity close to the feed to
the first stage. This new stream, F3 in Figure 11, is then treated in
the same manner and the process repeated until the desired amount of
glycerin is recovered.
The values of rejections, fluxes, and membrane areas that are
used in the multistage treatments are shown in Table 24.
TABLE - 24
Multistage Stage No. % Rejection Flux,
Model G P I gfd
Membrane Area
Sq. Ft.
Forward Feed
Countercurrent
2, 4, 6
1A,1B
Forward Feed 1, 3, 5
Countercurrent 2A,3A,2B,3B
33 66
22
78
100
78
15
5
4
300
1440
1440
G, P, and I stand for glycerin, polyglycerine, and inorganics. The
material balance shown in Figure 11 reflects the performance obtain-
able with this system. Addition of water between each stage is
necessary to maintain favorable stream viscosity. However, the de-
sign of the recovery unit is based on reclaiming most of the added
water along with the valuable constituent,glycerin. The three units
shown indicate the decreasing trend in the recovery of glycerin with
52.
-------
IjO
W,P,G,I : Ibs./min
G
El
Fi W:T0.4
W:~98.6 P: 1-2
P: 15.3 G: 0.7<
G: 2.6 I: 0.2
I: 2.6
t <
W: 34
FR KJ.
W: 64.6 W: 58.
P: 15.3 P: 14.
G: 2.6 G: 1
I: 2.6 I: 2
/
1)
2
1
9
4
E
W:
P:
G:
I:
1
W:
2
25.0
0.2
0.3
0.0
R2 E
W: 15.4 W:
P: 1.0 P:
G: 0.4 G:
I: 0.2 I:
F3 *
W: 98.5
P: 15.1
G: 2.3
25
0
,3
"ZT0.5
1.2
0.7
0.2
(
R3
W: 5J
P: 11
G: ]
I: 2
E
W:
P:
"~ G:
I:
/
3)
5.0 ^
J.9
L'7 W:
2.4
4
25.0
0.2
0.3
0.0
P4
W: 1
P:
G:
I:
25
5.5
1.0
0.3
0.2
F5
W: 99.0
P: 14.9
G: 2.0
I: 2.6
(f
i
(
E6
W: 25.0
P: 0.2
*"G: 0.2
I: 0.0
. ..- -
E5 —
W:~40.5 p1 1^'^
P: 1.2 :;•"
G: 0.6 ' _'
I: 0.2
5)
3^ K-)
W: 58.5
P: 13.7
G: 1.4
I: 2.4
MULTISTA3E FORWARD FEED OPERATION
FIGURE-II
-------
the addition of more units. The schematic representing this model in
Figure 11 is an open system to leave the choice of installing addi-
tional units open as per the dictation of process economics and pollu-
tion standards for an individual application. In an actual plant
installation the final raffinate Rg , would be recycled. The raffinate
R^ could also be recycled in part to improve the recovery of glycerin
and concentration of the final raffinate.
Multistage Countercurrent Operation:
Figure 12 shows the multistage counter current system. The reaf-
finate from the first battery of cells is fed to the second identical
battery for further glycerin recovery. Additional units may be added
depending on the amounts of glycerin to be recovered. The values of
rejections, fluxes and membrane areas selected for various stages of
this model are listed in Table 24. Material balance on each battery
is obtained by solving algebric equations governing component flow
rates and the rejection characteristics of membranes. The weight
fraction of a solute in the product from any stage is assumed propor-
tional to the mean raffinate concentration (mrc) of the solute, where,
concentration of the solute (entering the stage +
leaving the stage)
mrc = - = - fi - = -
The following equations result from the stagewise material
balance of the system.
Stage - 1:
e - e- r = 0 (3)
- 1/2 (l-REJlc)
X /
Where c represents an individual component (water, glycerin,
polyglycerin , inorganics) of the stream. Equation (4) obviously
results from the mrc assumption listed above. Similar equations may
be written for Stage 2 and Stage 3.
= 0 (5)
. 0 (6)
54.
-------
W,P,G,I : Ibs./min.
J1A
IB
Ul
w
F P
RA G
W: 64.6 I
P: 15.3
G: 2.6
I: 2.6
R2A
W: 72.4
P: 15.8
G: 2.6
I: 2.8
R1A
7.3
0.7
0.3'
0.2
i
f
FTA
W: 104.:
P: 16.5
G: 3.'
I: 3.(
•\
W: 34
*x
\
\
I
1
I
? i
i
)
\
1
f
t
W:
P:
>
G:
I:
W:
P:
G:
k I:
XNx
I
W: 24.6
P: 0.4
G: 0.5
I: 0.0
E2A
"3T.9 W:
1.1 P:
0.8 G:
0.2 I:
FRB =
E3A
32.4
0.9
0.5
0.2
R3A
W: 74.0
P: 14.9
G: 2.1
I: 2.6
R1B
7.6
0.5 ,
0.2
0.2
R3A -
\
W:
. t
ROT
TB
W: 114. (
P: 16.'
G: 2.
I: 3.-
\
* ^
\
%
i
3
* -
3
I
W:
P:
G:
I:
\
i
t
2B
W:
P:
G:
•I:
81.5
15.5
2.3
2.8
34
i
\
\
W
P
G
I
W: 24.8
P: 0.4
G: 0.3
I: 0.0
E2B
32.4
0.9
0.5
0.2
E3B
~37.3
1.0
0.5
0.2
R3B
W: 83.2
v P: 14.5
G: 1.8
I: 2.6
MULTISTAGE COUNTERCURRENT OPERATION
FIGURE - 12
-------
Stage - 3:
r2c - 63c - r3c - °
W is added to the right hand side of equation (7) when expressing
water balance on Stage 3.
- 1/2 (1 - REJ) [ -^ + ~^ } = 0 (8)
Also, for the i stage extract rate (E ) and raffinate rate (R.)
are defined as:
i iw ia ip il
R. = r. + r. + r. + r... (10)
i iw ig ip il
Where e's and r's represent component extract and component raf-
finate rates. Subscripts W, G, P, I, as usual stand for water,
glycerin, polyglycerin and inorganics respectively. Thus 24 material
balance equations result for a 3-stage countercurrent battery which
when solved simultaneously yield individual component flow rates in
extract and raffinate streams from each stage shown in the Figure 12.
It is seen that even in this model the amount of glycerin recovered
(in E-^ and E^g) shows a decreasing trend. Consequently, the glycerin
recovery cost will have to be balanced against the cost of additional
membrane units.
56.
-------
Economic Analysis:
Each potential plant would require its own economic analysis.
Preliminary economic estimations may, however, be made based on the
mass balance, obtained for the three models discussed. The estimate
is worked out in detail for the multistage forward feed model as an
illustration. Results for the other two models, obtained identically,
are listed in Table 25. Engineering economics texts (11, 12) are
referred to for the estimation procedures.
Multistage Forward Feed Operation
Investment;
2 ?
Membranes and modules, initial cost, 5200 ft @ $25/ft $130,000
Pumps, 3-10 horsepower pumps rated at 15 gpm @ 600 psi 3,600
Total Investment $133,600
(assume no additional distillation cost; piping ignored;
controls included in above)
Operating Expenses (Annual basis):
2
Membrane Maintenance $2/ft 10,400
Manpower 1/4 man @ $8,500 for 4 shifts plus
.8 shifts to cover vacation, holidays, etc. 10,200
Utilities Power $70, Steam and Water free 100
Total Operating Expenses $20,700
Glycerin Credit:
0.8 Ibs/minute or 415,000 Ibs/year $0.15/lb. 62,250
Depreciation 10-year base 13,360
Total Credit ........................ $75 , 610
Payout :
Payout is calculated on the basis of the annual return diminished
by a 48% tax rate, plus the allowable depreciation (10 year basis).
No provision has been made for interest costs, since these can be in-
cluded in the risk analysis calculations. For examples, the capital
requirement of $133, 600 is paid back by the sum of the depreciation $,
$13,360 plus the net return per year (100 - 48%) ($41,550) . or
133,600 0 o
= 3'8 years
- (0.52) x (41,550) + 13,360
Rate of return = 26%
57.
-------
TABLE -25
ECONOMIC ESTIMATION
Single Multistage Multistage
Stage Forward Feed Countercurrent
Total Investment 33,600 133,600 163,600
Total Operating Expenses 12,650 20,700 23,300
Glycerin Credit 46,650 62,250 62,250
Payout Time, Years, Months 1 Yr.,7 Mos. 3 Yrs.,9 Mos. 4 Yrs.,6 Mos,
% Rate of Return 62.5 26 22
It is observed from Table 25 that complete removal of inorganics
raises the total investment appreciably and, therefore, results in a
much lower rate of return. The economics is thus very sensitive to
the level of removal of inorganics.
The cost of neutralization of the waste stream is not included in
this estimation since many plants may have some dispensable neutralizing
streams. For the particular stream studied in this report the pH of the
feed stream varied between 8 and 9. Based on pH 9 of the waste stream,
the cost of neutralization with phosphoric acid is estimated to be less
than $2,000/year.
The predominant item in this analysis is the membrane investment
and the maintenance cost. We have deli erately chosen costs for these
parameters higher than the normally reported valued in the literature.
To cite an example, most analyses report a membrane maintenance cost
of $0.60 to $1.00 per square foot per year as against the $2.00 per
square foot per year assumed here. The life expectancy of membranes
in the glycerol system remains to be established, therefore, a conser-
vative figure was used in this cost summary. Similarly, the initial
investment cost was based on a small pilot unit, and should be less
for larger installations.
Major savings can be achieved in the process if the total membrane
area can be reduced. This study indicates that flux through the mem-
brane can be increased by increased turbulence, reduced viscosities,
and prefiltration to remove suspended matter and/or high molecular weight
species. Since the required membrane area is inversely proportional
to the flux, a 25% increase in flux would result in a similar reduc-
tion in the required membrane area. The payout in years would corres-
pondingly be reduced to about 2.9 years with a 34% for forward feed
operation and 3.4 years with a return of 29% for the countercurrent
operation. The data in this study indicate an improvement of this
magnitude or greater, and should be attainable with an additional
investigation.
58.
-------
Section VI
PROPYLENE GLYCOL WASTE STREAM (Dow Chemical Company)
During the period of the present grant, The Dow Chemical Company,
Freeport, Texas, also had an R&D grant from FWQA for the investigation
of various methods for disposal of a waste stream from their propylene
glycol production plant (Project No. 12020 EEQ). In establishing the
best disposal system for this stream, Dow and FWQA requested Gulf
South Research Institute to study reverse osmosis as one of the methods
having potential to aid in the solution of this problem. The waste
stream had the following composition:
NaCl % 9.3-10.4
Propylene Glycol ppm 500 - 2000
Propylene Chlorohydrin ppm 20 - 70
Propylene Oxide ppm 100 - 500
It was indicated that it would be desirable to reduce the NaCl
concentration to less than 1% (in the product) while increasing the
propylene glycol concentration to greater than 1%.
Several problems were apparent for the application of reverse osmo-
sis technique to this stream. First, the propylene glycol and NaCl
molecules do not differ appreciably in size. In those systems where
there is a large difference in the average diameter of the solutes, the
separation by reverse osmosis is relatively simple. However, as the
solutes approach the same size, the task becomes more formidable.
Second, the high salt content of this stream further complicates the
problem due to a very high osmotic pressure in a system with semiper-
meable membranes. Separation under these conditions can occur only if
preferential interaction of solutes in the system occurs within the
membrane phase. Based on this hypothesis,several membranes were ex-
amined for their ability to effect separations.
Both flat plate cells, where a number of candidate membranes
could be screened, and the tubular module described in the preceding
sections were used for evaluation. The data are presented in Tables 26
and 27. The results show that among these membranes the maximum
selectivity was offered by the membrane G-23; however, even this was
far from adequate. This may be due to the fact that when a membrane
is selected with low enough NaCl rejection to reduce the osmotic
effect to an acceptable level, pore flow becomes a dominant mode of
transfer and, therefore, very little separation of solutes is observed.
The preliminary tests thus showed that reverse osmosis did not
appear to be a very promising technique for the treatment of this
stream. Further investigation was, therefore, discontinued.
59.
-------
TABLE - 26
DOW WASTE STREAM - I
MEMBRANE
MEMBRANE GEOMETRY
PRESSURE
ANALYSIS
Membrane
G-16
G-17
G-19
G-20
G-21
G-22
G-23
G-24
G-25
Flux
gfd
54.5
35.1
29.5
41.1
21.1
23.5
10.5
35
12.4
INORGANIC
Feed Prod.
Weight % Weight %
9.1
9.1
9.2
9.2
10.0
10.0
10.1
10.1
9.4
8.6
7.7
7.3
8.0
7.9
8.0
6.5
8.3
5.1
: CELLULOSE ACETATE (EASTMAN)
: FLAT SHEET
: 800 psi
: DOW CHEMICAL COMPANY, FREEPi
Rej. %
5.5
15.4
20.7
13.0
21.0
20.0
35^6
17.8
55.7
Flux
gfd
54.5
35.1
29.5
41.1
21.1
23.5
10.5
35
12.4
ORGANIC
Feed Prod
ppm ppm
1600
1600
1800
1800
1610
1610
1890
1890
2040
1600
1610
1440
1680
1650
1560
1640
1680
1202
Rej. %
0
0
20
6.7
0
3.1
13^2
12.5
41
-------
TABLE - 27
DOW WASTE STREAM - II
MEMBRANE
MEMBRANE GEOMETRY
PRESSURE
FLOW RATE
Test Feed Composition
(1)
10.2% Sodium Chloride
876 ppm Propylene Glycol
110 ppm Epichlorohydrin
200 ppm Propylene Oxide
CELLULOSE, GSRI, B-112
TUBULAR
600 psi
0.85 gpm
Hours of
Operation
Flux
gfd
7.5
Rejection
%
9.5
5.5
61.9
10.2% Sodium Chloride
876 ppm Propylene Glycol
110 ppm Epichlorohydrin
200 ppm Propylene Oxide
7.5
9.7
11.3
90.0
36.0
(1)
Analysis performed by Dow Chemical Company, Freeport, Texas,
The chemical stability of propylene oxide and epichlorophdrin
make the results questionable for these compounds.
61.
-------
Section VII
OTHER WASTE STREAMS
It was beyond the scope of this grant to investigate many of the
industrial petrochemical waste streams for which reverse osmosis would
be a potential treatment technique. In the early phases of this pro-
gram, however, several solutes which represented classes of pollutants
were screened to delineate areas where reverse osmosis technique would
find potential applications. Of these, glycerol and polyglycols have
already been discussed in this report. A disperse dye (Cibacet RED 3B)
was chosen as another one of the model or representative solutes. The
compound has a molecular weight of 239 and a basic structure as shown
below:
This three-ring compound was selected to represent chlorinated hydro-
carbons used in the production of insecticides and herbicides, unsatu-
rated hydrocarbons such as substituted stilbenes and ethylene oxide
derivatives in a reverse osmosis system. This model has a limited
solubility in water and it tends to agglomerate. The results of the
screening program are shown in Table 28.
In all cases excellent rejections were observed. The flux with
the synthetic streams approached that of distilled water indicating
little, if any, fouling of the membrane. The glycerin recovery system
illustrates various problems encountered in an industrial application
and each waste stream must, therefore, be evaluated for problems due
to viscosity, membrane fouling and the presence of other solutes. The
results indicate that this technique should be applicable to obtain
excellent separations. The economics of such an operation, however,
will have to be worked out for each individual system.
Further investigation of this stream was discontinued due to the
lack of industrial cooperation and time.
62.
-------
TABLE - 28
DISPERSE ORGANIC DYE SYSTEM
MEMBRANE SOURCE : UNIVERSAL WATE
(EXCEPT GSRI E
MEMBRANE GEOMETRY : FLAT SHEET
SOLUTE CONCENTRATION : 1000
Membrane
No.
GSRI B-112
IV- 6 4- 2
IV-64-2(81)
IV-64-3
IV-64-3
IV-64-3 (81)
IV-64-4
IV-64-4(81)
IV- 6 4- 5
IV-64-5(76)
IV-64-5(81)
CAB-171-15
CAB-171-25
Membrane
Type
C
CAB
CAB
CAB
CAB
CAB
CAB
CAB
CAB
CAB
CAB
CAB
CAB
Pressure
psi
600
600
800
600
800
600
800
600
800
600
800
600
800
600
800
600
800
600
800
600
800
600
800
600
800
ppm
Flux
gfd
7.5
61
92.9
36.9
48.4
54.1
58.3
55.8
62.9
22.4
30.2
36.6
43.4
20.5
25.4
32.2
34.9
47.2
63.5
19.5
27.5
7.15
8.66
6.81
9.42
Rejection
89
90
97.3
> 99
> 99
96
97.3
98.9
> 99
> 99
> 99
94.2
97.2
> 99
> 99
93
94.2
> 99
> 99
> 99
> 99
> 99
> 99
98.2
> 99
63.
-------
Section VIII
ACKNOWLEDGMENTS
The financial support of the Louisiana State
Science Foundation in the form of a grant, matching the
Water Quality Office, Environmental Protection Agency
grant for this project, is acknowledged with sincere
thanks.
Mr. William T. Hackett, Jr., Executive Director,
Louisiana State Department of Commerce and Industry,
and Mr. Vernon Strickland, the Project Administrator,
provided the necessary administrative assistance for the
project.
Mr. Shyamkant Desai performed laboratory experiments,
did calculations, evaluated the collected data and wrote
an initial report. Mr. James K. Smith and Dr. Elias Klein
directed project activities and helped prepare the final
report. Dr. Robert E. C. Weaver of the Department of
Chemical Engineering, Tulane University, New Orleans,
made many valuable suggestions.
Shell Chemical Company, Norco, Louisiana, cooperated
in our efforts to investigate typical industrial effluents
for reverse osmosis. Mr. Robert Trautner of Shell was
particularly helpful in several related discussions.
Support of the project by the Water Quality Office,
Environmental Protection Agency, and the valuable sug-
gestions provided by Mr. Lawrence Lively and Mr. James Horn,
the Grant Project Officer, are acknowledged with sincere
thanks.
64.
-------
Section IX
BIBLIOGRAPHY
1. Eller, J.R., Ford, D.L., Gloyna, E.F., J. of AWWA, 149 (Mar.1970).
2. Agrawal,J.R., Sourirajan, S., Ind. Eng. Chem. Process Des. Develop.,
2, No. 1 (January 1970).
3. Kimura, S., Sourirajan, S., A.I.Ch.E.J., JL3, 497(1967).
4. Desai, S.V., M.S. Thesis, Department of Chemical Engineering,
Tulane University, New Orleans.
5. Shreeve, N., Chemical Process Industries, McGraw Hill Publications,
564, ed. 1968.
6. American Standard Product Information Manual for the Pilot Unit.
7. Office of Saline Water, Research and Development Progress Reports,
Nos. 143, 339.
8. Brian, P.L.T., Ind. Eng. Chem. Fundam., _4, 439(1965).
9. Dresner, L. , J. Phys. Chem., 69, 2230(1965).
10. Tien, C. and Gill, W.N., A.I.Ch.E.J., 12, 722(1966).
11. Peters, M. , Plant Design and Economics for Chemical Engineers,
McGraw Hill Publications, New York, 1958.
12. Perry, J.H., Chemical Engineers' Handbook, McGraw Hill Publications,
New York, 4th edition, Section 26.
65.
-------
Section X
APPENDIX
CHEMICAL ANALYSIS OF GLYCERIN WASTE STREAM
Glycerin Determination:
A 3 ml sample was added to a small beaker along with 25 ml
deionized water. The pH was adjusted to 6 using 0.5N HC1. The
contents of the beaker were then poured through a 50 ml buret filled
with 30 ml of Amberlite IRA-400 resin in the chloride form. The
resin was washed with 100 ml deionized water. The sample and all
washings were collected in a large Erlenmeyer flask. The pH of this
sample was brought to between 3 and 4 using 0.5N HC1. One drop of
Dow Corning Antifoam B was added and nitrogen was bubbled through the
sample for 15 minutes. Then 3 drops of 0.5% methyl red indicator was
added and the sample was neutralized with 0.1N NaOH. 50 ml of 5%
sodium periodate was added, the flask was stoppered and allowed to
stand for 10 minutes. Next, 5 ml of 50%v ethylene glycol was added
and the flask was allowed to stand for 5 minutes. Three drops of
methyl red indicator were added and the solution was titrated with
0.1N NaOH to the end point.
A blank was run for each batch of sodium periodate used.
Note: All compositions referred to in this section are percent by
weight unless otherwise mentioned.
Calculation:
<* n • (A)(B)(9.21)
/. Glycerin = ^
where,
A = Volume of 0.1N NaOH required to reach end-point,
minus the blank.
B = Normality of NaOH used.
C = Volume of the original sample used.
D = Specific gravity of the original sample, by hydrometer
(range: 0.9 - 1.2).
Water Determination:
Duplicate samples, approximately 5 grams each, were weighed on a
Mettler balance in previously tared porcelain crucibles. The crucibles
were then placed in an oven at a temperature of 110°c for a minimum
of 4 hours. The samples were then removed from the oven and placed
in a desiccator for 1 hour. The samples were reweighed.
66.
-------
Calculation:
% Water = ^ ~ ^ x 100
where,
X = Weight of the crucible plus the weight of the original
sample.
Y = Tare weight of the porcelain crucible.
Z = Weight of the crucible plus the weight of the sample
after being heated to 110°c for 4 hours.
Inorganic Determination:
The samples used for water determination were then placed in a
muffle furnace at 700°c for 4 hours. The crucibles were allowed to
cool considerably in the muffle furnace before being placed in a
desiccator for 1 hour. The samples were then weighed.
Calculation:
% Inorganics = -j- -y x 100
where,
Q = Weight of the crucible plus the weight of the original
sample used in the water determination.
R = Tare weight of the crucible.
S = Weight of the crucible plus the weight of residue after
ignition in the muffle furnace
Polyglycerine Determination;
Knowing the percent by weight composition of the other three
components in the system, the percent by weight composition of poly-
glycerines was found by difference.
Calculation:
% Polyglycerines = 100 - (G + W + I)
where,
G = % Glycerin.
W = % Water.
I = % Inorganics.
67.
-------
Table 29 illustrates the typical reproducibility of the analysis
duplicate samples analyzed concurrently. Variations in day-to-day
operations lead to the larger deviation quoted on page 13.
TABLE - 29
REPRODUCIBILITY OF ANALYTICAL RESULTS
Sample % H20 By % H20 % Inorganic % Organic
# Weight Loss Karl Fisher
DS 24 F 86.94 87.35 1.77 11.29
86.73 1.78 11.49
FS 24 P 98.17 96.85 .10 1.73
98.24 .10 1.66
RE 5 F 88.97 86.45 1.48 9.55
89.04 1.49 9.47
RE 5 P 97.79 98.95 .08 2.13
97.80 .08 2.12
68.
-------
Access/on Number
Subject Field & Group
SELECTED WATER RESOURCES ABSTRACTS
INPUT TRANSACTION FORM
Organization
GULF SOUTH RESEARCH INSTITUTE, New Orleans, Louisiana.
Title
POLYMERIC MATERIALS FOR TREATMENT AND RECOVERY OF PETROCHEMICAL WASTES
10
Authors)
Klein, Elias
Desai, Shyamkant V.
Smith, James K.
Weaver, Robert E. C.
16
Project Designation
EPA Grant No. 12020DQC
21
Note
22
Citation
Final Report, October 1970, Water Quality Office, Environmental Protection Agency.
68 pages, 12 figures, 29 tables, 12 references.
23
Descriptors (Starred First)
*Reverse Osmosis, *Membranes, *Separation-Techniques, *Water Pollution Treatment,
Effluents, Neutralization, Viscosity, Pressure, Economic Evaluation.
25
Identifiers (Starred First)
Flux, Rejection, Recovery, Concentration, Multistage Operation.
27
Abstract
Reverse osmosis is used as a unit operation to study the utility of polymeric
membrane materials for petrochemical plant waste treatment. A glycerin plant waste
stream containing glycerin, polyglycerines and inorganics was treated using
cellulose acetate butyrate memoranes. These membranes offered best performance
in the pressure range 600-800 psig. The data, obtained on a semi-pilot scale
equipment, was used to project a plant scale operation. It is shown that such a
recovery is possible, using single or multistage designs with an attractive
return on investment. The report also includes preliminary studies on disperse
dye and propylene glycol plant waste stream. (SVD).
Institution
WR:I02 (REV. JULY 1969)
WRSIC
SEND TO: WATER RESOURCES SCIENTIFIC INFORMATION CENTER
U.S DEPARTMENT OF THE INTERIOR
WASHINGTON. D. C. 20240
* GPO: 1969-3= 5-339
------- |