United States
Environmental Protection
Agency
Industrial Environmental Research
Laboratory
Cincinnati OH 45268
EPA-600'2-79-209
December 1979
Research and Development
&EPA
Advanced
Filtration of
Pulp Mill Wastes
-------
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.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-79-209
December 1979
ADVANCED FILTRATION OF PULP MILL WASTES
by
John T. McKinnon
International Paper Company
Mobile, Alabama 36601
Grant No. R803667-01-1
Project Officers
Donald L Wilson and Ralph H. Scott
Food and Wood Products Branch
Industrial Environmental Research Laboratory
Cincinnati, Ohio 45268
Industrial Environmental Research Laboratory
Office of Research and Development
U. S. Environmental Protection Agency
Cincinnati, Ohio. 45268
-------
DISCLAIMER
This report has been reviewed by the Industrial Environmental Research Laboratory, U. S.
Environmental Protection Agency, and approved for publication. Approval does not signify that
the contents necessarily reflect the views and policies of the U. S. Environmental Protection
Agency, nor does mention of trade names or commercial products institute endorsement or
recommendation for use.
-------
FOREWORD
When energy and material resources are extracted, processed, converted, and used, the
related pollutional impacts on our environment and even on our health often require that new and
increasingly more efficient pollution control methods be used. The Industrial Environmental
Research Laboratory Cincinnati (IERLCi) assists in developing and demonstrating new and
improved methodologies that will meet these needs both efficiently and economically.
This report is a summary of laboratory and pilot plant studies of reverse osmosis (hyper-
filtration) and ultrafiltration of pulp mill waste waters using dynamically formed (or dynamic)
membranes. Although the results obtained were not sufficient to provide scale-up data for such
systems, the studies did provide useful information that will aid the progress of future work.
The operating experience with the pilot units should be especially beneficial to the design of
future systems, as most of the difficulties encountered, both mechanical problems and membrane
problems, could be eliminated with redesigned equipment. This report will be valuable to
researchers interested in the treatment of industrial waste waters with membrane systems. Tubular
dynamic membrane systems have the potential of greater flexibility and operating efficiency under
more extreme conditions than the conventional precast membranes. Persons concerned with in-
plant treatment of pulp mill waste waters should be especially interested in this study, as the pilot
plant operating data are indicative of the removal efficiencies of the dynamic membrane units
now appearing on the market. For further information, please contact the Food and Wood
Products Branch of the Industrial Environmental Research Laboratory, Cincinnati, Ohio.
David G. Stephan
Director
Industrial Environmental Research Laboratory
Cincinnati
iii
-------
ABSTRACT
This report covers the experimental development and the pilot plant operation of
dynamically formed reverse osmosis and ultrafiltration membranes for the treatment of highly
colored waste waters. The dynamic membranes were formed on porous supports typically having
pores in the submicron range. The dynamic membranes were made of substances added to solu-
tions circulated under pressure past the supports. During the pilot plant testing, membranes were
supported by porous ceramic and carbon tubes in an outside flow arrangement. A hydrous
zirconium (JV) oxide-poly aery lie acid membrane was used in the reverse osmosis mode to treat a.
pulp mill decker filtrate. A hydrous zirconium (IV) oxide-silicon (IV) oxide membrane was used
in the ultrafiltration mode to treat a bleach plant caustic extraction filtrate (Ej).
During the pilot plant operation, the reverse osmosis unit treated the decker filtrate to a
reusable quality and the ultrafiltration unit was able to concentrate the bleach plant waste water
to a solids concentration of greater than 10%. Both units achieved better than 98% color removal
and high levels of organic compound removal. The ultrafiltration unit was capable of fluxes of
4.07 m3/m2/d (100 gfd) or higher, and the reverse osmosis unit produced fluxes in excess of
1.63m3/m2/d(40gfd).
The membranes used in this pilot plant study, which are not commercial membranes, showed
a relatively high degree of instability and variability. Mechanical and design problems precluded
the gathering of enough data for a scale-up or an economic evaluation. More development work
is required to eliminate the operational problems and to determine the membrane life.
This report was submitted in fulfillment of Grant No. R803667-01-1 by International Paper
Company unde/ the partial sponsorship of the U. S. Environmental Protection Agency. This
report covers a period from August 1,1976 to April 28,1978.
IV
-------
CONTENTS
Foreword iii
Abstract iv
Figures vi
Tables ix
Abbreviations, Symbols, and Metric Conversions x
Acknowledgments xi
1. Introduction 1
2. Conclusions 2
3. Recommendations 3
4. Membrane Development 4
5. Experimental 9
Laboratory studies 9
Reverse osmosis 9
Ultrafiltraton 27
Summary 38
Pilot plant studies 39
Equipment descriptions 39
Operating experiences 45
Membrane formation procedures 50
Cleaning of units 50
Forming the membranes 50
6. Results 53
Reverse osmosis unit 53
Ultrafiltration unit 61
7. Discussion of Results 83
Ultrafiltration unit 85
Water recovery 87
Economic evaluation 87
References 89
Appendices 93
A. Principles of Operation 93
B. Membrane Supports 97
C. Analytical Procedures 99
-------
FIGURES
Number Page
1 Flow diagram of high-circulation-velocity hyperfiltration loop 10
2 Hyperfiltration of simulated pulp-washing effluents:
washing and regeneration tests 11
3 Effect of filteraid pretreatment of used 19-tube 0.27 y Selas ceramic bundle . ... 12
4 Effect of silicate on hyperfiltration by hydrous ZR(IV) oxide-polyacrylate
membranes on 0.27 y Selas ceramic tubes 13
5 Comparison of hyperfiltration of simulated pulp-wash liquor: ceramic and
filteraid-treated porous metal supports 15
6 Hyperfiltration of simulated unneutralized pulp-wash liquors
with various membranes 18
7 Hyperfiltration of unneutralized simulated kraft-wash liquors 19
8 Hyperfiltrationoof diluted weak black liquor 21
9 Hyperfiltration of simulated kraft pulp-washing effluent:
Mott porous stainless steel tubes and Selas ceramic tube 23
10 Hyperfiltration properties of hydrous Zr{IV) oxide-polyacrylate membrane
on individual Selas tubes in 19-tube bundle (two tubes broken) 24
11 Pressure drop versus fluid flow for Selas module with two 19-tube
bundles in place 25
12 Effect of prefiltration of feed on hyperfiltration of pulp-washing effluent 26
13 Effect of oxygen on hyperfiltration of simulated pulp-wash liquors r 28
14 Hyperfiltration of kraft-pulping wash effluents: detergent regeneration'; 29
15 Flow diagram of ultrafiltration loop 30
vi
-------
16 Ultrafiltration of bleach plant effluent by hydrous Zr(IV) - Si(lV)
oxide membrane on 19-tubeSelas bundle 31
17 Ultrafiltration of bleach plant effluents with hydrous Zr(IV) - Si(IV)
oxide membranes 32
18 Ultrafiltration of kraft bleach plant effluent by hydrous Zr(IV) - Si(IV) oxide
membrane on Mott porous stainless steel tubes and Selas ceramic tube 33
19 Comparison of Ultrafiltration of bleach plant effluent by several
dynamic membranes 34
20 Comparison of Ultrafiltration and kraft pulping of effluents. Hydrous Zr(l V)
oxide membrane on 19-tube Selas ceramic bundle 37
21 ORNL-IP Co. hyperfiltration loop schematic 40
22. Pilot plant installation 41
23. Hyperfiltration module assembly containing eight 19-tube bundles:
four assemblies in rack 42
24 Close-up of reverse osmosis unit 43
25 IP Co. Ultrafiltration loop schematic 44
26 Selas ceramic tube bundles . .> 46
27 Broken tube bundle with moderate fiber accumulation 47
28 Tube bundle installation: reverse osmosis unit 49
29 Run H-1. Reverse osmosis treatment of decker filtrate 54
30. Run H-2. Reverse osmosis treatment of decker filtrate 57
31 Run H-3. Reverse osmosis treatment of decker filtrate 59
32 Run U-1. Ultrafiltration of caustic bleach effluent (E!) 63
33 Flux vs. total solids in concentrate, Run U-1 65
34 Pressure drop vs. tola I "sol ids concentration. Run U-1 67
35 Run U-2. Ultrafiltration of caustic bleach effluent (Et) 69
36 Run U-3. Ultrafiltration of caustic bleach effluent (El). Continuation of
Run U-2 with new membrane and tubular supports 71
vii
-------
37 Run U-4. Ultrafiltration of caustic bleach effluent (Ej),
carbon tube module only 73
38 Run U-5. Ultrafiltraton of caustic bleach effluent (Ej) using
ceramic tube bundles 76
-.
39 Run U-6. Ultrafiltration of caustic bleach effluent (E}) using
carbon tube module 80
VIII
-------
TABLES
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
Summary of Tests of Mott Porous Stainless Steel Filters
Hydrous Zr(IV) Oxide-Polyacrylate Membrane on Raypak Ceramic Tube,
Pretreated With Filteraid
Ultrafiltration of Bleach Plant Effluents by Membranes, Dynamically
Formed of Kelco Biopolymers
Reverse Osmosis Treatment of Decker Filtrate, Run H-1
Reverse Osmosis Treatment of Decker Filtrate, Run H-2
Reverse Osmosis Treatment of Decker Filtrate, Run H-3
Ultrafiltration of Caustic Bleach Plant Effluent, Run U-1
Flux vs. Concentrate Solids Concentration, Run U-1
Pressure Drop vs. Concentrate Solids Concentration, Run U-1
Ultrafiltration of Caustic Bleach Effluent, Runs U-2 and U-3
Ultrafiltration of Caustic Bleach Effluent Using Carbon Tube Module,
RunU-4
Ultrafiltration of Caustic Bleach Effluent Using Ceramic Membrane
Support, Run U-5
Ultrafiltration of Caustic Bleach Effluent Using Carbon Membrane
Support, Run U-6
Summary of Reverse Osmosis Data
Summary of Ultrafiltration Data
Ultrafiltration Data, Individual Tube Bundles, Run U-5,
December?, 1977
Page
14
16
36
55
58
60
64
66
66
70
74
77
81
84
86
86
IX
-------
ABBREVIATIONS, SYMBOLS, AND METRIC CONVERSIONS
ABBREVIATIONS
BOD5
COD
TOC
%F»
gfd
m3/m2/d
Pa
psi
psig
m3/s
gpm
m/s
ft/sec
ppm
c.u.
ym
SYMBOLS
HN03
H2S04
J
NaOH
NaNO3
NaCI
NaHCO3
Na2CO3
Na2SO4
Zr(IV) - PAA
Zr(IV)-SiOV)
Zr(NO3)4
biochemical oxygen demand, 5-day test
chemical oxygen demand
total organic carbon
percent rejection of constituent
gallons per square foot per day
cubic meter per square meter per day
Pascals (Newtons per square meter)
pounds per square inch
pounds per square inch gauge
cubic meter per second
gallons per minute
meters per second
feet per second
parts per million
color units
micrometers (microns)
bleach plant caustic extraction filtrate
nitric acid
sulfuric acid
flux rate
sodium hydroxide
sodium nitrate
sodium chloride
sodium bicarbonate
sodium carbonate
sodium sulfate
hydrous zirconium(IV) oxide polyacrylic acid
hydrous zirconium(IV) oxide silicon(IV) oxide
zirconium nitrate
METRIC CONVERSIONS
MULTIPLY BY
gfd
psi
ft/sec
gpm
atm
4.074x10
6.8947572 x103
3.048 X10"1
6.3088 x10"s
1.0135x105
TO OBTAIN
m3/m2/d
Pa
m/s
m3/s
Pa
-------
ACKNOWLEDGMENTS
The project was performed by International Paper Company, with Oak Ridge National
Laboratory participation under subcontract OIC-SNN-29. Oak Ridge National Laboratory is a part
of the U. S. Department of Energy and is operated by Union Carbide Nuclear Corporation. The
assistance of Dr. J. S. Johnson, Jr., Dr. A. J. Shor, Mr. Gary Westmoreland, and Mr. Philip Hayes
of Oak Ridge National Laboratory in performing the laboratory studies, constructing the pilot
units, and providing technical support is gratefully acknowledged. The project manager for
International Paper Company was Mr. Lea J. Breithaupt, Jr.
Special appreciation is extended to Mr. George Bartholomew, manager of the Flotrortics
Division, Selas Corporation of America, for providing the ceramic tube bundles used as membrane
supports and forhis technical assistance.
The cooperation and assistance rendered by the Moss Point Mill maintenance and laboratory
staff throughout this project are gratefully acknowledged.
XI
-------
SECTION 1
INTRODUCTION
The highly colored waste waters from integrated pulp and paper mills are the subject of great
concern from the standpoints of both control technology and impact on receiving streams. The
removal of color from these waste waters is difficult and expensive because the bulk of the color
is in the form of soluble organic compounds. A number of color removal processes are available,
each with its advantages and disadvantages. Coagulation processes are effective, at the expense of
sludge handling problems. Destructive methods, such as ozone oxidation, produce effluents of
high BOD content. Adsorption processes, such as activated carbon or resin adsorption, pose the
problem of the regeneration of the adsorption media. Conventional biological treatment can be
effective in removing low levels of color from some waste; however, biological treatment is
inadequate for consistently removing the high levels of color from a pulp and paper mill waste.
Reverse osmosis and ultrafiltration are high energy and equipment intensive processes, but they
have unique advantages that make them attractive methods for treating pulp mill waste waters.
Reverse osmosis and ultrafiltration have the capacity to selectively remove compounds with
different molecular weights and to produce clean permeates for recycle. In a pulp mill, especially
a bleached-pulp mill, several waste streams are fairly low in volume and quite high in color concen-
tration, and thus they contribute heavily to the total mill effluent color load. Reverse osmosis and
ultrafiltration can be applied successfully to such waste streams. This is accomplished by recycling
the high molecular weight compounds which contribute to the color load into the mill recovery
system and producing clear permeates for reuse or disposal.
The earlier membrane systems used for this purpose have been limited in application because
of the pH and temperature extremes encountered in pulp mill waste waters, the low flux rates ob-
tained, and problems with membrane fouling. Dynamically formed membranes of hydrous metallic
oxides have been found to give good results with pulp mill and textile mill waste waters. These
membranes can be operated at pH and temperature extremes that would adversely affect con-
ventional membranes while at the same time obtaining much higher flux rates.
This report is the culmination of a pilot plant test project in which dynamic membranes were
used to treat pulp mill waste waters in situ.
-------
SECTION 2
CONCLUSIONS
The pilot plant operation was erratic because of mechanical problems and design faults. It did
not produce enough data for successful scale-up calculations. However, the project did identify
several important facets of dynamic membrane reverse osmosis and ultrafiltration treatments of
waste waters. The use of dynamic membranes may be a viable alternative to conventional
membrane systems. When the units were running well, high fluxes and good rejections of
dissolved organics were obtained. However, the high degree of membrane instability demonstrated
throughout the pilot plant work revealed a need for further development before dynamic mem-
branes will be commercially acceptable.
In several of the experimental runs, difficulties were encountered in membrane formation
because of mechanical problems, which caused poor results in waste water operation. It was
demonstrated in subsequent runs that a poor membrane could be improved by a combination of
washing techniques and by applying extra layers of membrane. Low fluxes during operation with
a good membrane could be raised through appropriate washing procedures.
The ultrafiltration unit membranes could successfully be stored for long periods of time if
imme.sed in water. This indicates that individual modules can be prepared and stored for future
use, such as with conventional membrane systems.
-------
SECTION 3
RECOMMENDATIONS
The present regulations on the discharge of highly colored waste waters indicate that color
removal processes will be necessary in the future. As pulp mills become more involved with internal
recycling, the waste streams will become lower in volume and more concentrated than they are
currently. The present high cost of membrane systems may be offset m the future by the need to
treat such waste streams. Thus, the development of dynamic membrane systems should continue,
especially as these systems are more amenable to pulp mill waste waters than conventional
membrane systems.
Specifically, more emphasis should be placed on the mechanical design of the units themselves.
Prefiltration of the pulp mill wastes should be performed before membrane treatment to lessen
fouling and clogging by pulp fibers. Proper design of the units can make fiber clogging less critical
than it was with these pilot units. The future trials should focus on using an inside flow arrange-
ment similar to a shell and tube heat exchanger. Porous stainless steel tubes would allow the use of
this configuration at higher pressures than with ceramic tubes. The steel tubes would also allow for
welded tube assemblies, and any defective or broken tubes could be plugged in a manner similar to
the procedure used on boiler tubes. This would allow a module with individual broken tubes to
remain in service. Because the tubes would be supported on the permeate side, no flow restrictions
would exist to trap fibers.
Thus, the dynamic membrane system with an extensively prefiltered feed for outside flow
modules (carbon, ceramic, or steel tubes), or'a slightly prefiltered feed for inside flow, could be
developed for pulp mill wastes if care is taken to eliminate the operational problems encountered
during this project.
-------
SECTION 4
MEMBRANE DEVELOPMENT
With the advent of stricter limits on effluent quality, increased emphasis has been placed on
color removal processes. As a result, a number of processes have been developed, each with its ad-
vantages and disadvantages. Lime, alum, or other coagulants remove color effectively, but they
generate large volumes of sludge resulting in a disposal problem. Adsorption with activated carbon,
synthetic resins, and other adsorbents will also remove color from pulp mill waste waters. However,
disposal of the spent carbon or of the regeneration streams complicates the integration of this pro-
cess into regular mill operations. Biological oxidation in conventional secondary waste treatment
systems is not consistently adequate and oxidation with ozone, hydrogen peroxide, or other oxidants
is too expensive or otherwise difficult to justify. Other processes, such as solvent extraction, amine
precipitation, and electrochemical removal are known but not yet attractive on a commercial scale.
Membrane processes, such as reverse osmosis (also called hyperfiltration) and ultrafiltration,
have been applied to many situations in which separation of a solute from a solution is desired (see
p. 39 for more detailed description). The processes first became practical in the early 1960's with
the development of the first membranes. Since then they have been used to desalinate seawater
and brackish groundwaters for potable uses, treat municipal and industrial waste waters, and re-
cycle numerous contaminated process solutions which would otherwise be discarded. In the field
of industrial waste water treatment, the potential for application of membrane processes is great
because they allow separation of possibly valuable by-products and reuse of clean water. Although
the processes are energy intensive, they generate no sludges and do not consume a large amount of
chemicals.
MEMBRANE CHARACTERISTICS
The growing interest in reverse osmosis and ultrafiltration for desalination and waste water
treatment has generated the need for improved membranes. Thus, over the past decade a number of
membrane materials have been used in attempts to overcome the shortcomings of the original cellu-
lose acetate membranes. The different types can be subdivided into two groups: precast and
dynamic membranes.
Precast Membranes
The use of reverse osmosis and ultrafiltration for commercial scale applications became possi-
ble with the development of cast cellulose acetate film as a membrane material. Although other
membranes have been developed in attempts to improve performance, cellulose acetate is still
widely used (6, 9,19, 21, 23, 24, 29, 30, 31, 32). It can be cast in the form of flat sheet, spiral-
wound, or tubular modules. It is limited in that pH and temperature must be limited to ranges of
3-8 and up to 35-38°C, respectively. At pH values outside this range the ester groups hydrolyze, and
-------
the membrane structure deteriorates at higher temperatures. The membrane gives flux rates of 0.6 to
0.8 m3/m2/d (15 to 20 gfd) at ultrafiltration pressures of 2.5x10s to 1x106 Pa (35 to 150 psi) and
flux rates of about 0.3 m3/m2d (7 gfd) at reverse osmosis pressures of 2x106 to 4x106 Pa (300 to
600 psi). At these conditions, cellulose acetate will reject molecules of molecular weights of 10,000
or higher when operating in the ultrafiltration mode and molecules of molecular weight of 80 or
higher when operating as a reverse osmosis unit.
To improve flux rates and increase removals, various modifications to the basic cellulose
acetate structure have been tried. Cellulose acetate is a very polar molecule which accounts for the
passage of water and polar molecules through the membrane. Cellulose acetate butyrate (CAB) and
cellulose acetate triacetate (CAS) were developed by substituting butyl and acetyl groups, respec-
tively, for free hydroxide groups in the basic cellulose acetate structure (8). The rejections im-
proved, somewhat, at the expense of lower fluxes. Other modifications, such as cellulose acetate
0-propyl sulfonic acid (CAOPSA) and cellulose acetate (E398-10), have also been used, with little
or no improvement over the original structure (25).
The aromatic polyamide membranes, developed by Du Pont, are generally formed into hollow
fiber modules (9, 21, 29, 31). The polyamide material can be operated in a pH range of 4-11 at
temperatures up to 38°C. It gives similar removals and fluxes as cellulose acetate and thus is useful
when pH values higher than 8 are encountered. One drawback is that the membrane deteriorates in
the presence of chlorine, which precludes its use as a cleaning agent. The IMS-type membranes,
made of polyethylenimine crosslinked with tolylene 2,4- diisocyanate, were developed by the North
Star Research Company (9,12, 25, 31). This material is essentially nonpolar and is usually cast in
tubular modules. The NS membranes (NS-1, IMS-100, and NS-200) give better removals of polar
organics than cellulose acetate, and they have been used successfully at very low and high pH values
(0.5,12 respectively) (25). The flux rates are quite similar to those obtained with cellulose
acetate membranes.
Polysulfone membranes have the advantage of operating at higher temperatures (up to 88°C)
and wider pH ranges (1-13) i(14, 23, 28, 32, 33). Solute removals are slightly lower than those
usually obtained with cellulose acetate, but the membrane can be used to treat effluents which
could not be treated with cellulose acetate. Fluxes are higher than those obtained with cellulose
acetate but drop rapidly as the total dissolved solids concentration increases. Polysulfone mem-
branes are usually cast in tubular form and are sometimes used as a support for other membranes.
Several other membrane materials have been used for reverse osmosis and ultrafiltration, but
they have not gained as widespread application as those mentioned above. Union Carbide
Corporation developed an ultrafiltration unit using a proprietary inorganic membrane (UCARSEP")
which can be used in pH ranges of 1-14 and temperatures up to 130°C, cast on a tubular carbon
support (22,32). .Membranes cast of poly 2,2 (m-phenylene)-5,5 bibenzimidazole and sulfonated
polyphenylene oxide have been developed, but they gave much lower removals of salts and organics
than cellulose acetate, polyamide, and IMS-type membranes (31). A membrane formed from poly
(ether/amide) deposited on a polysulfone spiral-wound module obtained 99+% removal of salt and
fluxes slightly belter than cellulose acetate membranes at 6.9x106 Pa (1000 psi). This membrane
can be used in a pH range of 3-12 and temperatures up to 55°C, and it resists chemical attack, mak-
ing cleaning easier than with other membranes (34). \ Another new membrane was developed by
coating a polysulfone hollow fiber with a furan resin and crosslinking it with sulfuric acid (35). The
membrane gives excellent salt rejection (99+%) at pressures of 5.5x106 Pa (800 psi) and can operate
-------
in pH ranges of 1.5 to 13. The flux rate obtained is low, in the order of 0.08 to 0.12 m3/m2/d
(2-3 gfd), but productivity can be high in a hollow fiber module because thousands of fibers can be
used in a single module.
A number of membranes have been developed for reverse osmosis and ultrafiltration which
can be used in various configurations and in more extreme conditions than cellulose acetate.
However, the flux rates and solute rejections have not improved dramatically with the precast
membranes. Fouling problems have been lessened but not eliminated. It is anticipated that more
effort will be expanded in the future in an attempt to produce better membranes.
Dynamic Membranes
In addition to the developmental work being done with precast membranes, it was found that
membranes could be formed dynamically by circulating |a suspension of a suitable substance over a
porous support under pressure. The substance coats the porous support to form a membrane layer
with a porosity less than that of the support material. Such a membrane is called a dynamically
formed, or dynamic, membrane. The first feasible dynamic membranes were formed at the Oak
Ridge National Laboratory in the mid1960's. In 1966 it was reported that a membrane made of
hydrous thorium oxide had been developed (35). It was found that other metals, capable of form-
ing hydrous oxides, such as Fe(lll), Zr(IV), Bi(lll), and Sn(IV), could also form dynamic mem-
branes. Studies in later years showed that dynamic membranes could also be formed from natural
polyelectrolytes such as humic acid (37) and constituents usually found in pulp and paper mill
effluents (38,39, 40).
The metal oxide which has received the most attention is hydrous zirconium (Zr[IV]) oxide
because of its ion exchange capabilities. The fluxes obtained were high, even with thick mem-
branes, and it was discovered that a dynamic membrane gave better flux rates than a precast mem-
brane of the same material. It was also found that the dynamic Zr(l V) oxide membranes had a
self-rejecting property, in that after a certain thickness had formed, the membrane material would
begin to reject further formation. The dynamic membranes were first applied to desalination of
seawater and brackish waters, but it was found that they would reject organic solutes as well as
salts. In tests with spent sulfite liquors, it was found that a Zr(IV) oxide dynamic membrane
formed before application of the liquor gave better removals than membranes formed with the
liquor alone, or a mixture of the liquor and zirconium (36). A separate study showed that dynamic
membranes could be formed from calcium based spent sulfite liquor, but performance was very
erratic. The optimum performance obtained 90% rejection of color at flux rates of 1.2 m3/m2 /d
(gfd) at 3.5x106 Pa (500 psi) (39).
In 1969, a dual layer membrane consisting of an underlayer of hydrous Zr(l V) oxide and an
overlayer of poly acrylic acid (PAA) was found to have ion exchange characteristics which indicated
a number of possible applications. The first membranes were formed on porous carbon tubes
manufactured for use as electrodes, but in later work other porous supports, such as ceramic and
stainless steel tubes and polymeric materials, were successfully used. It was found that the best
pore diameter was approximately 0.25 ym, and that a precoating of a filteraid would be necessary
for supports with pore diameters greater than 10 ym. Concentration polarization was found to be
more important with dynamic membranes than with precast types, so it was initially believed that
high circulation velocities (greater than 6.1 m/sec [20 ft/sec]) would be needed to obtain good
flux rates and rejections. High pressures were found to be necessary for good rejections with the
Zr(IV) - PAA membranes. Tests with salt solution showed that, at a pressure of 6.6x10s Pa(95
-------
psi), a flux rate of 27.1 m3/m2/d (666 gfd) and 19% rejection were obtained, while at a pressure
of 6.6x106 Pa (950 psi) the flux dropped to 6 m3/m2/d (147 gfd) with a 94% rejection (36). This
lowered flux rate was still much higher than those being currently obtained with precast membranes.
Tests using Zr(l V) - PAA membranes to treat textile wastes were reported in 1973 (41). At a
pressure of 6.9x106 Pa (1000 psi) and a temperature of 60°C, 99% rejection of color was obtained
at flux rates of 1.6'3.2 m3/m2/d (40 to 80 gfd). The tests showed that high velocities did indeed
alleviate fouling and also that higher operating temperatures resulted in higher fluxes. Four
different porous supports were used: ceramic tubes (0.27 ym pore diameter) manufactured by Selas
Corporation of America-Flotronics Division, Union Carbide carbon tubes (563-6C), and two steel
tubes wrapped with polymeric sheets (Millipore, 0.22 ym pore diameter, and Acropor AN, 0.45 ym
pore diameter). Further tests were also being performed using lignins to form a dynamic membrane
on a cellulose acetate membrane that had deteriorated (38). The results indicated that a dynamic
membrane formed from pure lignin and/or pulp and paper mill effluents could "tighten" an old
membrane or close up small defects in a leaking module. This lignin membrane was not dislodged
under high velocity flow conditions but was easily removed by regular cleaning techniques, such as
detergent washes. However, the membrane had to be formed at a pH of 4 to be effective and would
only last for 3 or 4 days.
The Office of Saline Water contracted with Oak Ridge National Laboratory to build a pilot
plant for the purpose of studying dynamic membrane desalination. The tubular supports available
were nearly equal in quality, so ceramic tubes manufactured by the Ferro Corporation were
arbitrarily chosen. These tubes required precoating with diatomaceous earth before membrane
formation because the pore diameters were approximately 10ym. The flow velocity was studied,
and it was found that reducing the velocity from the usual 9.1 m/sec (30 ft/sec) down to 3 m/sec
(10 ft/sec) did not significantly affect the flux rate. The percent rejection was reduced slightly,
but the pressure drop across the test module was reduced from 3.2x10s Pa (47 psi) to 4.4x104
(6.4 psi), respectively, allowing significant savings in pumping costs. The formation of the zircon-
ium layer was formed from an initial value of 200 m3/m2/d (5000 gfd) down to 40.7 m3/m2/d
(1000 gfd). If the zirconium layer were allowed to build up further, the final flux after the poly
acrylic acid layer formation would be too low to be practical. The performance of the membrane
was found to be adversely affected by hard water, and pretreatment to soften the water was
necessary. In actual pilot plant tests, it was found that reducing the flow velocity from 7.1 m/sec
(30 ft/sec) to 4.6 m/sec (15 ft/sec) resulted in a slight decrease in percent rejection because of
concentration polarization. However, the pressure drop across the sixteen tubular modules
(mounted in series) decreased, and the flux actually improved in the last few bundles (36).
Dynamic membranes for reverse osmosis and ultrafiltration were also being tested for treating
pulp and paper mill effluents in the cooperative efforts between Oak Ridge National Laboratory
and International Paper Company (40). Using several membrane supports, the following tests were
performed: (1) reverse osmosis treatment of simulated screen room effluents and caustic bleach
plant waste water using Zr-PAA membranes and (2) ultrafiltration treatment of the caustic bleach
plant waste water with zirconium and polyvinylpyrrolidine membranes. The Zr-PAA membrane,
operated at 6.9x106 Pa (1000 psi), gave 99+% color and 95-98% organic carbon rejections at flux
rates of 1.3 to 3.9 m3/m2/d (31 to 96 gfd), while treating the screen room effluent. Treatment of
the bleach plant effluents gave similar removals, but the concentrated feed had so much chloride in
it that it would be impractical to introduce it into the mill recovery system. The ultrafiltration unit
-------
(single layer zirconium membrane) gave 99% color and about 90% total organic carbon removals.at
operating pressures of 1.4x106 Pa (200 psi). When the pressure was raised to 6.9x106 Pa (1000
psi), the chloride rejection and resulting chloride concentration in the concentrated feed was too
high, as with the reverse osmosis unit. Similar results were obtained with the poly(vinyl pyrroli-
dine).
Later work with kraft mill wastes involved testing of a number of membrane materials, using
bleach plant waste waters (42). The original Zr-PAA membrane was the best choice for reverse
osmosis applications; whereas the poly(vinyl pyrrolidine) was the favorite for the ultrafiltration.
However, for ultrafiltration, three other membranes also looked promising: the Zr single layer
membrane; a polymer called pluronic 98 (80% poly[oxyethylene], 20% polytoxypropylene])
manufactured by Wyandotte Chemicals Co.; and a dual layer of hydrous zirconium and silicon
oxides. An investigation involving textile dyeing waste waters compared dynamic Zr-PAA mem-
branes with precast cellulose acetate and aromatic polyamide (26). The dynamic membrane gave
equal removals, while operating at higher temperatures and wider pH range feeds with no prefiltra-
tion. The dynamic membrane also obtained higher flux rates and allowed higher productivity.
Dynamically formed membranes have been shown to be capable of producing high flux rates
and rejections of various solutes. Presently, they are not used as widely as the precast types, simply
because they were developed later. However, the engineering aspects of the design of waste water
treatment systems for industrial effluents, such as temperature, pH, pretreatment needs, and
reliability, make dynamic membrane systems quite attractive. As others have recognized, in the
design of membrane systems to remove high molecular weight compounds, the choice of
engineering is more important than the choice of the membrane (8).
8
-------
SECTION 5
EXPERIMENTAL
LABORATORY STUDIES
Studies performed at Oak Ridge National Laboratory have established that dynamic membrane
reverse osmosis and ultrafiltration are capable of removing color from pulp and paper effluents (40,
41). Additional laboratory studies on operating parameters were necessary to obtain data for pilot
plant work.
Reverse Osmosis
An extended study with a 19-tube (ceramic) bundle in a flow scheme as shown in Figure 1 was
performed and is summarized in Figure 2. The bundle had been used previously for textile dye
wastes at temperatures up to 94°C and was cleaned by immersion in a chromic acid solution before
formation of a hydrous Zr(IV) oxide-polyacrylate membrane. The membrane gave a salt rejection
of 94% and a flux rate of 3.1 m3/m2/d (75 gfd) at the end of membrane formation. The waste
water fed to the module was a simulated decker filtrate produced by diluting weak black liquor
(nominally 12-15% total solids) to a concentration of about 1% total solids. The flux dropped to
about 1.6 m3/m2/d (40 gfd) in approximately 1 week, at which time the unit was washed with a
0.1M_ Na2C03 NaHC03 solution (pH =3) which restored performance. A second wash at about
250 hours was less successful, and a subsequent wash was performed at a lower pressure, in the
hope that the lower flux would make the wash more effective. Another wash sequence at about
340 hours included washes with 0.1 JVl HN03 and 0.1 JW H2SO4, but they had little benefit. The
bundle was washed with acid and base, and the membrane reformed, but the flux deteriorated
rapidly and improved only slightly after a 0.1 JW H2S04 wash. The bundle was removed, soaked
in chromic acid to remove the previous membranes, and a new membrane formed. The flux rate
dropped to 1.2 from 1.6 m3/m2/d (30 to 40 gfd) and stabilized at about 1.2 m3/m2/d, maintaining
this level for the duration of the run.
In an attempt to determine whether a change in the pore dimensions on the surface accounted
for the deteriorating properties seen in Figure 2, the tube bundle was cleaned again and pretreated
with a filteraid before the next membrane formation. The pretreatment did not cause a definite
improvement, as seen in Figure 3. The inadvertent shutdowns prevented a clear conclusion, because
membranes formed on filteraid layers tend to be unstable if operating interruptions occur.
Ultrafiltration tests have indicated that the presence of silicate helps to slow flux decline. This
same bundle was tested with hydrous Si(IV) oxide present in several different patterns during
membrane formation. The results are shown in Figure 4 (one run was with a single tube rather than
a bundle). However, little if any improvement was obtained in cases where Si(IV) was present
during the membrane formation.
-------
PRESSURE
CONTROL
VALVE
Figure 1. Flow diagram of high-circulation-velocity hyperfiltration loop.
10
-------
250
300
350
1
X
i-
tcp
tut
uj
CE
400
100 p
"
80 -
60 -
40 -
20 -
100 -
80 -
60 -
40 -
A
100 r-
50 '-
40C
f
I
I i
i g
£ 4
I 5
B U
i- s
* ' *?
s
u 5
i ?
7 *v
* ,
)
^
N b '
;
!;
!"]_!
II
^ 5
; §
i i
\t
% t
^
\ ,
l
450
(
_
. , . \ , . , . \
I
f
O
i
* °«
... i .... i1
. .1 . I .... 1
500 550
_
9
* ^
o"
s
U
I
§1- . . 1 . . . . 1 . . . . 1 . . . . 1 . . . , -
-i -j
O j
"1s
si
Jr
-^ .
.-... ,V .,....!.,.. i .,...
.
A n r i iii i
600 650 700 750 80
4.0
3.0
1 n
I >U
0
OPERATING TIME (hrs)
'Total organic carbon
\ Shutdown; t Fresh Diluted Weak Black Liquor Introduced
Figure 2. HyperfMiration of simulated pulp-washing effluents:
washing and regeneration tests,
(950 psi; 15 ft/sec; 55°C; pH 7-9)
(6.55x106Pa;4.6m/sec)
11
-------
a
|
Ji
X
1
80
60
40
20
0
90
-50
10
100
50
n
DO ° o o 0
- o 0 -o
1 1 1 1 1
_
0
_
~o
o o o -5
o °
o
- 1 1 1 1 1
^**^
i / JL i it i 1
3.5
3.0
2.5
2.0^
1.5
1.0
0.5
f
0
20
40
HOURS
60
^ Shutdown
t Fresh Feed
Figure 3. Effect of filteraid pretreatment of used 19-tube 0.27^ Selas ceramic bundle.
(950psi; 15 ft/sec; 55°C ; pH 7.3-8.6 )
(6.55x106Pa;4.6m/sec)
12
-------
9 100
90
80
100,
CJ
t 50
o
a.
o>
20
10
0.05 I
/IS Nad
Q2
i i i i
J I
I I I
SIMULATED PULP WASHING EFFLUENTS
J L
1
< I I I
J L
1
I'll
J L
4.0
3.5
3.0
2.5
2.0
1.5.
1.0
0.5
0.5
10
1.0 5
HOURS
No Silicate
Silicate Layer from 5ppm
Silicate Layer from 50 ppm
Silicate Layer Formed on Hydrous Oxide
Before PAA Deposition (single tube)
50
100
}
19-Tube Bundle
Figure 4. Effect of silicate on hyperfiltration by hydrous ZR(IV) oxide-polyacrylate
membranes on 0.27/1 Selas ceramic tubes.
(950 psi; 15 ft/sec; 55°C; Feed: Weak Black Liquor Diluted to 1% Solids pH 7 to 8 3)
(6.55x1 06 Pa; 4.6 m/sec)
-------
Other membrane supports were tested in addition to the porous ceramic tubes. Tubes made
of porous 316L stainless steel were provided by the Mott Metallurgical Corporation (Farmington,
Connecticut). All but one were 6.35x10"3m (0.25 in) outside diameter with a thin layer of fine
pores on the outside. The listed manufacturer's pore size refers not to actual pore diameter but to
the particle size removed when operated with a cross flow of feed past the surface. As seen in
Table 1, the nominal pore sizes tested were 0.2 and 5 ym, however, the manufacturer's literature
suggests that the actual pore sizes may be as much as a factor of 10 greater than the nominal size.
The optimum pore size appears to be in the range of 0.1 to 0.5 pm, and earlier work indicates that
filteraid pretreatment is desirable for pore sizes greater than 1 ym. As seen in Table 1, the
membrane properties after formation confirm that these Mott tubes have pore sizes too large for
use without filteraid.
The membranes formed after a filteraid pretreatment gave operating properties within the
customary ranges. The filteraid pretreatments used in Runs 10Cl and 10CJ were fine diatomite
(Johns-Manville Celite 505) followed by an even finer carbon black (Cabot Corporation Regal
SR). Carbon black alone was used in Run 7NM and was fed with the zirconyl chloride. The
membranes from Runs 10-CI and 7NM were tested with the simulated decker filtrate (weak
black liquor diluted to about 1% total solids and neutralized). Initial rejection and fluxes were in
the normal ranges, but the membrane in Run 10Cl deteriorated dramatically overnight, pro-
ducing a black filtrate the next morning. The tube was backwashed, and the membrane for Run
10CJ formed. In Run 7NM, the performances of both the Mott and Selas tubes were stable
overnight (Fig. 5).
TABLE 1. SUMMARY OF TESTS OF MOTT POROUS STAINLESS STEEL FILTERS
~ 15 ft/sec; 25-30°C; 950 psig; pH ~ 4 (hydrous ZR (IV) oxide), 6.5-7 (PAA)
Run
No.
1O-CG
10-CH
10-CI
10-CJ
7-NK
7-NL
7-NM
Support,
Mfgs.
Designation
(1) Inside
(O) Outside
5y(l)
5y(l)
5 y(l)
5 y(l)
0.2 yd)
0.2 y(O)*
0.2y(l)
Hyd. Zr(IV) Oxide With PAA Layer
Filteraid
None
None
(boiled Zr(IV))
25 ppm Celite 505
25 ppm Regal SR
25 ppm Celite 505
25 ppm Regal SR
None
None
25 ppm Regal SR
(pre-mixed with
zirconyl chloride)
gpd/ft2 F
210
310
300
320
135
685
270
^obs. %
25
26
46
40
35
44
19
gpd/ft2
34
90
100
40
14
84
50
Robs- %
66
23
90
93
64
63
95
Feed
0.05 M NaNO3
0.05 M NaNO3
0.05 M NaNO3
0.05 M NaCI3
0.05 M NaCI
0.05 M NaNO3
0.05 M NaNO3
10 micron tube, with 0.2 y rated coating on outside.
14
-------
Q.
>d$
>^
. ""
100
90
^
*, 80
M
£ 70
60
50
~ 100
j en
2 ou
9
* "yr\
C\J
10
8
* ~ 2.
~>~ O
-. g Metal
- oo Ceramic,0.27u ° ~
1 , 1 , o 1 , 1 * ,
l-,^ Ceramic Metal ,
'« * o TOC
A sulfur A A -
H **"* a Conductivity
3 I*"0 »*o o g
ko
_ D _
O
1 , 1 , 1 1 1 1
i- Metal
oj o»» Q o Ceramic, 0.27«. -
8 ^^ . 0 _i
1 . e 1 i 1 . 1 * <>i ° -*
1 00 0 ° ° -I
' 1 , ° 1 , 1 . 1 ,
f 1 t Fresh Feed 1
1 , 1 . t , 1 , It,
) 10 20 30 40 5
TIME (hrs)
Figure 5. Comparison of hy perforation of simulated pulp-wash liquor:
ceramic and filteraid-treated porous metal supports.
950 psi 15 ft/sec 55°C
6.55x106Pa 4.6 m/sec
4
3 5
2 fc
I
|
0
-------
However, the next morning fresh unneutralized feed was introduced and the rejection from both
tubes fell, precluding a direct comparison with past experience. The conductivity rejection for the
Selas tube was 72% at a flux of 1.30 m3 /m2 /d (32 gfd) after 1 day, while that for the Mott tube
was 58% at a flux of 2.24 m3/m2/d (55 gfd). The total sulfur rejections were 91% for the Selas and
79% for the Mott tubes. At the time of this test, stainless steel tubes with pore sizes in the 0.5 ym
range were not available, and the use of a filteraid was necessary.
After the stainless steel tubes were removed, a ceramic tube manufactured by Raypak (now
Rev-0-Pak, Inc., Newbury Park, California) was installed. The tube was mounted in an outside flow
configuration. A filteraid pretreatment (25 ppm Celite 505 and 50 ppm Regal SR) was used prior
to membrane formation. The membrane deteriorated because of an unscheduled test loop shut-
down, and a second acidic PAA layer formation was performed. As seen in Table 2, the membrane
properties were close to the expected range. The 0.45 ym pore size Acropor control membrane
support gave 92% salt rejection at 3.63 m3/m2/d (89 gfd) using a circulation velocity of 9.1 m/s
(30 ft/sec) as opposed to 4.6 m/s (15 ft/sec) with the Raypak tube. This membrane was used as a
control to check general system operation. Neutralized simulated decker filtrate was introduced,
and conductivity rejection was good although the initial flux decline was high. After a day of oper-
ation, weak black liquor was added to bring the total solids content up to 8%. The flux and
rejection declined immediately and the clear, colorless permeates became turbid and yellow. At the
end of the run a tube inspection revealed that part of the black precoat had been lost. The Acropor
support also showed unusual deterioration, indicating that some upset had occurred. Its conduc-
tivity rejection was 59% at a flux of 1.92 m3/m2/d (47 gfd).
TABLE 2. HYDROUS Zr(IV) OXIDE-POLYACRYLATE MEMBRANE ON RAYPAK
CERAMIC TUBE, PRETREATED WITH FILTERAID
(15 ft/sec)
Tirno
Mrs.
Temperature Flux
pH °C gpd/ft2
Robs.%
Conductivity
After membrane formation (0.05 M NaCI)
0
1
3
6
23
24
27
47
6.6
Introduced diluted weak
7.3
-
7.5
.8
I ntroduced diluted weak
7
8.2
25
black liquor, 1% solids
53
55
55
55
black liquor, 8% solids
55
35
74
76
56
49
46
10
63
94
94
94
94
95
72
57
In the above results, as well as in past work with pulping effluents, neutralization of the feed
has been considered necessary. This was done because the polyacrylate layer of the membrane
tends to be displaced at pH values above 10 (rapidly above pH 11)? A lower pH was also thought
to be necessary for the rejection of the low molecular weight sulfur containing compounds (for
chemical recovery). However, the necessity for feed neutralization has been re-examined in the
light of several factors. It has been noted that the reaction of some feed constituents with the
sulfuric acid used for pH adjustment produces species which adversely affect membrane perform-
ance. Fouling of the membrane may also be less rapid at higher pH levels. Although rejection of
16
-------
NaCI by hydrous Zr(IV) oxide membranes is low in its cation-exchange mode at high pH, Na2S04
is appreciably filtered because of its divalent co-ion. Thus, permeates can be recycled into the
earlier stages of pulp washing and the sulfur compounds will eventually arrive in the chemical
recovery system. Selas Corporation has modified their ceramic tubes to be more resistant to hot
alkaline solutions, and tube damage from high temperature operation has been minimized. It is
possible that a dynamic membrane formed from feed constituents over the preformed layer may
have a useful rejection capacity. Thus, if the polyacrylate was degraded at the higher pH of the
unneutralized feeds, adequate performance may still occur. All of these observations indicated
that pH adjustment was unnecessary, and thus subsequent testing was done with unneutralized
feeds.
A run in which three membranes were simultaneously tested is summarized in Figure 6. The
three membranes were (1) a Kelzan (xanthan gum) membrane on a single Selas ceramic tube, (2) a
hydrous Zr(IV) oxide-polyacrylate membrane (Zr(IV)-PAA) mounted on a Selas 19-tube bundle,
which had been severely fouled in a previous test, and (3) a hydrous Zr(IV)Si(IV) oxide
membrane, also on a Selas 19-tube bundle that had been previously used in ultrafiltration tests
with bleach plant and other wastes. The Zr(IV)PAA membrane gave fluxes of only 0.81 to
1.22 m3/m2/d (20 to 30 gfd) in the previous test, but there was no significant deterioration in flux,
even though solids concentrations comparable with the level of weak black liquor were attained.
The feed in this test was the diluted weak black liquor, but without any acid addition for pH
adjustment. The unusual number of shutdowns indicated in Figure 6 were the result of tube fail-
ures mostly in the Zr(l V)Si(l V) oxide bundle which was constructed of thin-walled tubes and
had an extensive history of long operating periods at low pressure. This bundle was demolished
after about 40 hours in this run. The rejections for all parameters were higher for the Zr(l V)PAA
membrane than for the others. The most notable aspect was the relatively high total sulfur re-
jection by this membrane at high water recovery (84%) when the concentration of weak black
liquor (about 12% total solids) was reached. At this point, the permeate contained 2300 ppm total
organic carbon (TOO, 500 ppm total sulfur, and 360 units of color, while the concentrated liquor
contained 34,500 ppm TOC, 3400 ppm total sulfur, and 235,000 units of color.
The test results indicated in Figure 7 are for a newly formed hydrous Zr(IV) oxide-
polyacrylate membrane (88% salt rejection and flux of 4.07 m3/m2/d [100 gfd] at 25°C). The
fluxes for this membrane, which was formed on a new Selas 19-tube bundle, were substantially
higher than those for the fouled membrane m Hgure 6. The rejections obtained were similar.
The next investigation was a direct comparison of the stainless steel tubes with the ceramic
tubes. The pore sizes of the stainless steel tubes were too large and an application of filteraid was
necessary to form a thin layer pf suitable pore size. The history of the five membranes is summar-
ized below, along with the symbols used to represent the results in Figure 8.
D A Mott tube with nominal pore size of 0.2 ym (which could be 10 times larger) was used
as the support for a Zr(IV)PAA membrane using Cabot Regal SR carbon black as a
filteraid. The rejection was 94% at a flux of 2.44 m3/m2/d (60 gfd) for 0.05 JM NaN03.
A Mott tube with a nominal pore size of 5 ym was used as the support fora Zr( IV) PAA
membrane with no filteraid, although some residual carbon from previous runs was
noticed. The rejction was 94% at a flux of 2.04 m3/m2/d (50 gfd) for 0.05 M NaN03.
17
-------
1UU
96
92
88
100
SO
80
~ff\
f\j
60
50
Af\
4O
80
60
40
20
"f
10
5
°0
i TT-
A 1_ 1 1 1 1
o o
A
B D
'x
~o 1 1 1 1 1
Q 0
1 A o AS
o- o
1 1 1 1 1
1 1 1 1 I ^1
5 10 15 20 25 '
Til
A,O,D TOC
*»» Color
I I i
A,o,a Conductivity
*,«, Sulfur
1 1 1
D
A
1 l«
t Feed Addition
1 Shutdown
H 1 1-4-
30 35 40
VIE (hrs)
] B o -!
A O
D C
| A | | |
t
o
o
O _
A A TJ
1 1 1 1 * ~
O
A
0 0 A -0
,. , ... ,^
s^ _
^r~rfi 1 1 i u =
45 50 55 60 6J
j
3
T3
2 ^
E
A Kelzan on Single Tube
o Hydrous Zr(EZ) Oxide-polyacrylate on 19-tube Bundle
a Hydrous Zr (12)-Si(12) Oxide on 19-tube Bundle
Figure 6. Hyperfiltration of simulated unneutralized pulp-wash liquors with
various membranes.
950 psi K ft/sec
6.55x1 06Pa 4.6m /sec
^ se^ Ceramlc
18
-------
KO j^iuu
° 6 £ 99
4 /"N/"\
2 lOO
g *- 80
0 in
g|60
J 80
X
3 eo
u_
x 12
Q.
1O
^< O **
^ yj
n
0 0
0
o
o
, 1 . 1 1 1 1 1 1
-2° ° o o0 °
; no A
3. D D
c n a
I 1 T A . i
5 o
0 o
o
0 0
1 1 1 1 1 . 1 1
p.. , ,,.»: . .
* i . i t ti . i .
V
1 ,
1
4-1
1 i 1
o TOC
A Total Sulfur
n Conductivity
i i i
i i i
0
o -
i . i i
o o o
A
A 0 ° _
o a K
i i i i
e
. , , , v-
0 _
_
75-80°C\~
\
5
4
3
2
I
10 20 30
40 50 60 70 80 90 100
TIME (hr$)
Figure 7. Hyper-filtration of unneutralized simulated kraft-wash liquors.
950 psi 15 ft/sec
6.55x106Pa 4.6m/seo
55 °C t Fresh Feed
19
-------
O A hydrous Zr(l V) oxide membrane was formed on a Selas 19-tube ceramic (0.27 ym
average pore size) bundle. At pH 4, rejection was 40% at a flux of fO.4 m3/m2/d
(255 gfd) for 0.05 MNalMO3.
A A hydrous Zr(IV) oxide-polyacrylate membrane was formed on a Selas 19-tube ceramic
(0.27 ym average pore size) bundle. Rejection was 85% at a flux of 3.46 m3/m2/d
(85 gfd) for 0.05JV[NaN03.
V A hydrous Zr(IV) oxide membrane was formed on a Selas 19-tube ceramic (0.27 ym
average pore size) bundle and exposed to a polyacrylic acid solution at pH 2 for
5 minutes.
All were installed in a loop for tests with diluted weak black liquor, but preliminary tests with
0.05JVj_NaN03 indicated considerable deterioration, as the results below indicate (% R, flux in
m3/m2/d):
D (34%, 285) (67%, 125) O (10%, 385) A (64%, 115) V (63%, 190)
The two hydrous Zr(l V) oxide modules (O and v ) were removed, and the others subjected to
an acid cycle containing polyacrylic acid. The rejections were restored, but there was evidence of
membrane instability on the stainless steel tubes. The two hydrous Zr(l V) oxide bundles were
reinstalled, and unneutralized diluted weak black liquor trials performed with the results
summarized in Figure 8.
The initial fluxes through the Mott steel tubes were high and their permeates were highly
colored. However, the fluxes decreased and rejections improved, apparently because of plugging
of tube imperfections by feed constituents. During the overnight operation, the hydrous Zr(l V)
oxide module failed and shut down the loop, causing most of the feed solution to be lost. The test
was resumed the next morning with fresh feed and without this module. The results of this experi-
ment indicate several notable trends:
1 . The membranes formed on the stainless steel tubes were not as stable as those on the
ceramic tubes. The initial fluxes and rejections were comparable with those of the
ceramics, but shutdowns and contact with the simulated decker filtrate caused marked
deterioration. The color and TOC in their permeates were higher than for any of the
ceramic tubes, with or without polyacr,ylate. The rejections obtained with the 0.2 ym
tube were considerably higher than those with the 5 ym tube.
2. The tube bundle with only the hydrous Zr(IV) oxide layer gave poorer TOC and color
rejections than the other ceramic bundles (before.it failed). The membrane exposed
briefly to the polyacrylic acid was initially only slightly different from the Zr(IV)
PAA membrane. Its rejections at the end of the run were slightly lower.
f
3. The fluxes remained favorable at high water recovery, which compares well with the data
in Figure 6. However, at a concentrate level of 12% total solids, total sulfur rejection had
declined to 69% for the Zr(l V) - PAA membrane, as compared with 85% in Figure 6.
The total sulfur rejections shown in Figure 7 under similar conditions are intermediate.
The reason for these differences is not clear because rejections at low water recovery
levels were comparable.
20
-------
TOTAL FLUX CONDUCTIVITY REJECTION COLOR
SOLIDS (gpd/ft2) REJECTION (/?0bs,%) REJECTION
(%) <*obs.*> (*obs,%> _
uu
96
00
80
60
40
60
20
60
40
20
10
0
C
U-I& U
"^ V A A
- 0
V
_ a ° v _
I , I , *
~ V^8 **§ a 0 o A v TOC A A ~
- A T Total Sulfur ° -|
s ° *
0 0 -i
- ° , I , I o
° , i , i , o 5
- v A 8 A -
- R ° A V 0
0 * a a
0 v A TT
- o ° *
I , I A T
[ _ ^
3
1
) 10 20 30
TIME (hrs)
D 0.2/iMott
0 5/iMott ( Hyd. Zr(I2)OxrPAA Membrane)
A 19 Tube Selas BundleJ
* 19Tube Selas Bundle (Hyd. Zr(WOx.-PAA Membrane*)
o :19Tube Selas Bundle (Hyd. Zr(I2) Ox. Membrane)
* PAA at pH 2 Only
i SHUT DOWN t FRESH FEED
Figure 8. Hyperforation of diluted weak black liquor.
(pH 11 -12; 15 ft/sec; 950 psi)
(4.6m/sec;6.55x106 Pa)
21
-------
It appears that the exposure of the membrane to polyacrylic acid is beneficial. The fluxes of
the Zr(l V)PAA membranes were essentially higher than those of the single layer Zr(IV) oxide
membrane. Thus, the use of a polyacrylate overlayer seems worthwhile.
During the course of the overall study, stainless steel tubes with a pore size rating of 0.5 ym
were developed by Mott Metallurgical. Test results comparing hydrous Zr(IV) oxide-polyacrylate
membranes with a new 0.5 ym Mott tube, a new 0.2 ym Mott tube, an old 0.2 ym Mott tube, and
a Selas ceramic tube are summarized in Figure 9. The 0.5 ym tube gave rejections similar to those
with the new 0.2 ym but at much higher fluxes.
The variations in performance because of the variations in membrane supports and pH were
not the only operating parameters studied. The hydrodynamic environment within the 19-tube
module is important in membrane formation because of its effect on concentration polarization
and fouling. The flow regime through the Selas module is complicated, and it is difficult to predict
the situation for tubes in different positions. However, by comparing rejections and fluxes from
individual tubes, it should be possible to detect gross differences. If the concentrated solution
velocity over different tubes varies greatly, concentration polarization should result in obvious
variations in permeate quality and flux.
In a test with a bundle, originally of 19 tubes but with two broken off, individual tubes were
sampled with different overall flow velocities (Figure 10). The membrane condition was monitored
over the period of the test by taking composite measurements. As seen in Figure 10, the repeated
tests show some membrane deterioration at 4.57 m/s (15 ft/sec),.but the concentration polarization
effects can still be seen with the lower velocities. The percent rejection is obtained by a comparison
of trie conductivities of the concentrate and the permeates. The arrows indicate whether the flow
entered or exited at the base of the bundle. At the highest velocity of 4.57 m/s (15 ft/sec) the
variations do not seem greater than the usual random scatter of data, whereas the variations at
1.22 m/s (4 ft/sec) seem significant. This test indicates that differences in the hydrodynamic vi
environment of individual tubes in the 19-tube bundles are not as great at 4.57 m/s (15 ft/sec).
The design of reverse osmosis systems and the estimates of energy consumption of the
processes require information on the dependence of pressure drop through the system components
particularly the membrane modules and on circulation velocity. Measurements of pressure drop
through a module containing two Selas 19-tube bundles are summarized in Figure 11. Tests at
different temperatures showed that the pressure drops were slightly, but probably significantly,
lower at high temperatures. The effect of circulation velocity on pressure drop was also noted in
previous studies showing that a lower velocity caused a lower pressure drop through a series array
of modules. Consequently, the flux from the last few bundles was higher because of the higher ap-
plied pressure. However, a balance between lower pressure losses and increased concentration
polarization effects must be reached. A circulation velocity of 4.57 m/s (15 ft/sec) seemed to be
optimal.
The effect of feed prefiltration was tested using a filter capable of removing particles with
diameters as small as 5 ym. Hydrous Zr(l V) oxide-polyacrylate membranes were formed on a Selas
ceramic (0.27 ym) tube, and unfiltered and filtered solutions were fed to it. The results, given in
Figure 12, indicate that no gains in flux or rejection were obtained with prefiltration. Results with
membranes formed on Millipore {0.025 ym pore diameter) and Acropor (0.45 ym pore diameter)
film supports are comparable with those obtained on the ceramic tube.
22
-------
COLOR TOC (ft/sec)
Q 0.5 /iSS 3/8 in. OD 15
A A 0.2 /* SS 3/8 in. OD (NEW) 15
100
95
0
s ^
H8
8
R£JECTION, /Tat
IVITY
§8 $
| TO
§ 60
_ 150
N
x *:
3>ioo
U. (L
~ 50
0
10
X
a 8
o 0.2/i SS V*» in. OD (OLD) 32
o 0.27 fi CERAMIC 0.21 in. OD 15
I 1 1 1 IA [ I
O
§ 2
*
D D
0
0
O
" O
o o-g a g
o-x>
_D *
(m/sec)
4.6
4.6
9.8
4.6
S3
o
S -
0
^
_
* "
A A A
*
A A
O
O
o
1 . 1 1 1 1 1
~O
5
0
10 20 30 40 50
OPERATING TIME (hr)
60
70
Figure 9. Hyper-filtration of simulated kraft pulp-washing effluent:
Mott porous stainless.steel tubes and Selas ceramic tube.
[950 psi(6.55x106Pa); 55°C]
23
-------
15 ft/sec : 98gpd/ft2 8ft/sec : 98gpd/ft2 8 ft/sec 98 gpd/ft2 4 ft/sec : 105gpd/ft2
4.6m/sec : 4.0m/d 2.4m/sec : 4.0m/d 2.4m/sec : 4.0m/d 1.2 m/sec ' 4.3m/d
Tobs
, 90%
Tobs
, 88%
obs-
15 ft/sec: 101 gpd/ft2 8-ft/sec: 98gpd/ft2 8 ft/sec: 99 gpd/ft2 4ft/sec! 105gpd/ft2
4.6m/sec: 4.1 m/d 2.4m/sec: 4.0m/d 2.4m/sec: 4.0m/d 1.2m/sec: 4.3m/d
/?obs , 89.5% /?obs, 88% tf0bs , 83.4% /tobi , 77%
Next day, after pH adj.
99gpd/ft2, 4.0 m/d 15 ft/sec : 99gpd/ft2 15ft/sec = 104 gpd/ft2 15 ft/sec = 109 gpd/ft2
jf 92.5% 4.6 m/sec : 4.0m/d 4.6m/sec : 4.2m/d 4.6m/sec : 4.4 m/d
Order of Tests: (1)
(2)
s , 86%
(3)
/fobs , 87%
(4)
Figure 10. Hyperfiltration of properties of hydrous Zr(IV) oxide-polyacrylate membrane
on individual Selas tubes in 19-tube bundle (two tubes broken).
( Composite Tests Before Individual Tube Measurements Given Above, and Just After, Below )
950 psi , 65.5 bar
25° C
pH 7.5 - 8.5
0.05 M NaCI
gpd /ft2
VOID
-------
1/m
50
FLOW
100
200
i i i
gpm
10
15 20 30 40 5060
20
10
v>
CL
Q.
O
LJ
or
CO
LJ
1
ft/sec
m/scc
n =
950 psi
-
30°C 24 hr later //
in
0.5
0.1
10
15 20 25
1
1.5
i
2
i
3
i
4
I
5
6
i i
78
FLOW VELOCITY
Figure 11. Pressure drop versus flow for Selas module with two
19-tube bundles in place.
( 1 gpm O 0.4 ft/sec )
( IJl/mO 0.032 m/sec)
25
-------
100
- 90
-------
The laboratory tests were made on simulated decker filtrates by diluting shipped-in weak black
liquor from a nominal 12 to 15% total solids to about 1% total solids concentration. A possible
difference in the laboratory and the actual in-plant feeds is the degree of oxidation of the
constituents. An attempt was made to detect such an effect by running a test with the simulated
decker filtrate in a two-stage sequence: the first with oxygen bubbled through the feed and the
second with no oxygen added. A hydrous Zr(l V) oxide-polyacrylate membrane was formed on a
Selas 19-tube bundle, which at the end of formation gave a rejection of 88% at 3.38 m3/m2/d
(83 gfd) with a 0.05 JVJ NaCI feed. Simulated decker filtrate was introduced to the unit and the
permeate was discarded. Fresh feed was added to bring the total solids content to approximately
3% during the first phase. After the oxygen was shut off, the flux and rejection seemed to increase
slightly as seen in Figure 13. This increase, however, is masked by the weekend shutdown which
occurred at approximately the same time that the oxygen was discontinued.
Other researchers have found enzyme detergents, ethylenediaminetetraacetic acid (EDTA),
and trisodium phosphate (TSP) to be useful in cleaning fouling layers from reverse osmosis and
ultrafiltration membranes. In tests with naval shipboard laundry wastes, it was observed that the
fouling by some feeds could be removed to a large extent by exposure to the synthetic solution
approximating the waste. This solution contains Triton X-100, Na2C03, sour, and kaolinite. Its
effect on membranes previously exposed to diluted weak black liquor is shown in Figure 14. The
washes appeared to be beneficial, but the improvement diminished in successive cycles.
Ultrafiltration
In the flow scheme shown in Figure 15, an extended test of a dual layer hydrous Zr(IV)
Si(IV) oxide membrane was performed using a Selas 19-tube bundle. The results of the test
with unneutralized bleach plant caustic extraction effluent (Et) are shown in Figure 16. Limits on
availability precluded the continuous introduction of fresh feed, but at intervals some permeate was
discarded and made up with equivalent volumes of fresh feed. Color and TOC removals were good,
and chloride was rejected to only a minor degree. This is desirable in this case to prevent recycle
and build up of chlorides in the mill chemical recovery system. The flux was still about 1.83 m3/m
1.83 m3/m2/d (45 gfd) after 4 months of operation. A dye-waste study performed during an
interruption, reduced the flux, and further exposure to a solution containing silica did not restore
it. A wash with a Na2C03 solution brought the flux back up to 1.63 m3/m2/d (40 gfd).
Tests were also performed with Mott porous stainless steel tubes using hydrous Zr(IV)Si(IV)
oxide membranes. The Mott tubes were rated as 0.2 and 5 ym and, like the Mott tubes used in the
reverse osmosis work, are rated by the particle size filtered rather than the pore diameter. A filter-
aid was used with the steel tubes. The rejection of the steel tubes was comparable to that of the
ceramic tubes (Figure 17), with the flux from the 0.2 ym tube falling between that of the single
ceramic tube and the ceramic bundle. The 5 ym steel tube flux was comparatively low. The
comparison between the steel and ceramic tubes is complicated by the different velocities, but the
results seem to indicate that a Mott tube of a smaller pore size would be suitable for ultrafiltration
of bleach plant effluents.
Newer Mott stainless steel tubes used for reverse osmosis tests were also used for ultrafiltration
trials. A hydrous Zr(IV)Si(IV) oxide layer was formed on the new Mott tubes (0.2 and 0.5 ym),
on an old Mott tube (0.2 ym), and on a fresh Selas ceramic tube (0.27 ym) for comparison. The
results of the experiments with unneutralized Ej are shown in Figure 18. Color rejections were
27
-------
1UU
z ^
g*>
UJ §
2 !* 95
* "t ' » 4
I
l«
« |0
t
o
1 ° o
o
. 1 . I I
100 -
0 -7 90
_^ ^*
o « 80
^ (t 70
CC ** en
^ 80
x~_ 60
3^ 40
u. &
* 20
0
(A
<0^
*-<7> Q
^
a
n A °
° a A
o
°l
o~
- ° 0
r °° "-o..,
i i i i I i i i
"" X^~ o on -»CUoff
r^^ it i ; t 1 i 11 1 i
0 50
TIME (hrs)
v
o
6
a
A
-y
O
ft Color
o TOC
I
a
_
1
-
-
-
Conductivity
Total Sulfur
I
o
o
i 1
if 3 day Shutdown
4 i I
J
-
~n O
^
-
2
1
100
Figure 13. Effect of oxygen on hyperfMiration of simulated pulp-wash liquors.
950 PSi "
6.55x106Pa
4.6m /sec
7_8
28
-------
s^i
01-
to
.
«, 60
^_
i
3
x
800
600
400
200
0
120
100
80
60
40
20
? lo
. (
".1,1,1,1,
I
"
1 4 1 1 1 , 1 1 1 1
_ O
t>
o
«
*
o
, 1 , 1 . 1 1 1 .
r
~ . i/ . i i i , i .
^m
u>
m
M«
in
₯,
§
g Kl
e
e
1,1,1
o
> , 1 , 1
o
>
o
*° 0
1,1,1
... , ...__
r
, ,(, ,
* Bo- B(
, 1,1,
g 1 . 1 ,
.c
f
0
3- o
(/)
i *,
(!>
UJ j^
u '
° ,/. ,
"" o ORG. CARBON -
* SULFUR
» TOTAL SOLIDS -
1 , 1 , 1 . 1
1 , 1 i 1 , 1
JC
a
o
m , 1 . 1 , 1
O
UJ
O
1 1 1 . 1 , 1 ~~
5.0
4.0
3.0^
E
2.0
1.0
0 20 40 60 80 100 120 140 160 180 200 220 24O 260
TIME (hrs)
Figure 14. Hyperfiltration of kraft-pulping wash effluents: detergent regeneration.
Hydrous Zr(EZ) Oxide-polyacrylate Membrane on 0.27/i Selas Tubes
Open Points , Single Tube ; Solid, 19-Tube Bundle.
(15ft/sec; 950psi: pH,7toa5; 55°C)
(4.6m/sec; 6.55x1(T6Pa)
-------
HEAT EXCHANGER
TO DRAIN
PUMP
DP: Differential Pressure Gauge ; P: Pressure Gauge ;
PR: Pressure Regulator; R: Rotameter ; T: Temperature Gauge ; V:Valve,
Figure 15. Flow diagram of ultrafiltration loop.
30
-------
||ooo
° o o
Aoo°°
AB Q-
,* . *? V
j§ 90
80
JTO D D
0 00
,
a
00 00°
I
a
00° o
0
o
1
D a
OOn c
D O
_ 0 0 °
°,° °
o
o Color
o TOG
1 , o 1 o , | . 6>\
JS
. o
20
-iV"ai*g-T~-*l A (*-
~t* 1 A~"t~^ . I
» Conductivity
Chloride
jy. 70
j| 50
30
ca-1
2.5
2.05
.3
60
o oooo-
I , I
O O O 00 O
I.I.I
*£
UJUJ
100
50
0^
. rt^^n>
20 40 60 80 100
OPERATING TIME (days)
120 140
Figure 16. Ultrafiltration of bleach plant effluent by hydrous Zr(IV)-Si(IV)
oxide membrane on 19-tube Selas bundle.
(ZOOpsi, 7ft/sec, pH 8.5 to 9.5 )
(1.38x10* Pa; 2.1 m/sec )
I Shutdowns A. Interruption for Laundry and Dye Waste Tests
B. Silica Coating Renewal
C. Na2C03 Wash
31
-------
t/>
REJECTION (
IUU
98
96
94
92
90
88
40
20
n
i 1 v
o 0 A o TOC
- » * COLOR A
0 Q
7° a fc
V
~0
A , 1 , S 1 , I , 1 , 1 , 1
^A n 0 * o CONDUCTIVITY £^_0
t^I » * CHLORIDE
i 1 , 1 . 1 . 1 I 1 i 1
t :
i "
-
i ~
1UU
80
CM
*-
ifc-
\
o
a 60
X
i
n- 40
20
.K 10°
[0^ 50
'£ o
-
__ A
-
a a
o o
- ° i * i
1 1 1 1 1 1 1
1 , 1 , 1 . ,
) 10 20 30
A
A
-3
A -o
A "^
D a - 2 E
a
a
0 0
o ; -1
i , i . i .i
^ i
i / ,t i , i . r
40 50 60 70
TIME (hrs)
SUPPORT
Stainless Steel (Mott)
° 0.2/t
0.27/z Ceramic (Selas)
A Single Tube
o 19-Tube Bundle
ft /sec m/sec
11.4 3.5
11.4 3.5
4.7 1.4
9.8 3.0
t FRESH FEED
Figure 17. Ultrafiltration of kraft bleach plant effluents with hydrous
Zr(tV)-Si(IV) oxide membranes.
200 psi .
1.38x106Pa»
32
-------
25«C 55 *C
REJECTION.'!*^
100
*>
80
100
98
94
§ 20
8 0
60
«T 40
(ft/see)
D 0.5 ft. SS, 3/8 in. 00 1 5
0 0.2 /t 55. 3/8 in. 00 ) 5
A A 0.2 /i SS. 1/4 in. 00 33
0 0.27^1 CERAMIC, 0.2 in. 00 7
1
1
1
Q
a
A
8
4 1 I L
1
0 20 40 60 80 100 120 140
OPERATING TIME (hr)
Figure 18. Ultrafiltration of kraft bleach plant effluent by
hydrous Zr(IV)-Si(IV) oxide membrane on Mott
porous stainless steel tubes and Selas ceramic tube.
33
-------
100
i
~S 90
80 -
»-D '-*»
A
A
~ , 1 i 1 i 1 i 1 . 1 . 1 . 1 . 1 i 1
-J
0
a
1 i 1
1
o_
I.
20 -
!_- 0
o
-
0
g
1,1,
o
6-.B
1.1,1,1,1,1,1
o
»- ,
, 1 , 1 '
^
160
^^
V 140
"2 120
a
~ 100
x
| 80
" 60
40
r>c\
\
B«4
1. :
1 . . A«.0 A
O H O 3t "* A
B QQ n A .
^ A ^~
K nO OOO^Q O^ff A A
O o 8*" O
-
fc l 1 , 1 . 1 , 1 , 1 , 1 , 1 , 1 , 1 , 1 1 1 l
6.0
5.0
4.0
3.0
2.0
1.0
20 40 60 80
IOO 120 140
HOURS
160
Figure 19. Comparison of ultrafiltration of bleach plant effluent by several
dynamic membranes.
( 200 psi , 60°C, pH ~9 , 0.27/t Selas Ceramic )
(1.38x106Pa)
Membrane
Support
19-Tube Bundle Hydrous Zr{ 12)- Si (IZ) Oxide '5.7 1.7
Single Tube Pluronic 25R2 e.e 2.0
Single Tube Pluronic 25R5 6.6 2.0
Single Tube Pluronic 25R8 6.6 2.0
. -
mlsac TOC Color
34
-------
greater than 98%, except for the old 0.2 ym Mott tube, and the flux rate for the Mott 0.5 ym tube
was greater than for the new Mott 0.2 ym tube. Unfortunately, the time limit prevented incorpora-
tion of any new Mott tubes into the pilot plant operation.
In an earlier section, this report states that a dynamically formed membrane of block copoly-
mers of polyethylene oxide-polypropylene oxide appeared promising for the ultrafiltration of
bleach plant effluents. The polymer tested earlier was part of a class marketed as Pluronics by the
Wyandotte Corporation. Later trials with other members of this group were undertaken using Selas
ceramic tubes, and the results are shown in Figure 19. The designations in the key of Figure 19 are
from the manufacturer's literature. They refer to the molecular weight of the propylene oxide
portion of the molecule (25 is equal to molecular weight 2500), the position of the polypropylene
oxide structures on the molecule chains (R), and the approximate weight percentage of the poly-
ethylene (5, for example, stands for 50%). The code seems to include considerable approximation,
but the membranes listed in Figure 18 show a range of hyrophilic properties.
The Selas 19-tube bundle (with a hydrous Zr(IV)Si(lV) oxide membrane) was the same one
used for previous ultrafiltration tests. The inorganic membrane generally gave higher rejections
than the Pluronics. The fluxes from the organic membranes were higher initially, but tended
toward the same values as the inorganic membrane obtained at the end of the run. There was no
clear trend of flux or rejection variations when polyethylene oxide content was increased. The
increase in flux through the hydrous Zr(IV)Si(IV) oxide membrane was probably caused by
a recovery from exposure to highly fouling tannery wastes used in previous tests.
In addition to the Pluronics polymeric membranes, samples of biopolymers marketed by the
Kelco Corporation were tested. They have properties somewhat different from the other additives,
as the Kelgins are alginic acids. One Kelco biopolymer designated MV is of intermediate molecular
weight as indicated by viscosity, and another designated XL is of low molecular weight. Kelzan is
a xanthan gum with a molecular weight of about 2 million. The material used was more highly
refined than the grade used to enhance oil recovery. Kelzan is interesting because its chains are
said to possess unusual stiffness, and little change in the viscosity of its solutions because of salt
concentration or temperature can occur. The results of the ultrafiltration tests with these additives
are summarized in Table 3. The organic polymers had been originally formed on single Selas
ceramic tubes for reverse osmosis tests from a 0.05 IM NaCI solution. The hydrous Zr(lV) Si(IV)
oxide membrane was on a Selas 19-tube ceramic bundle used in previous ultrafiltration tests. The
data in Table 3 were taken after only 3 hours of operation but are similar to those obtained after
overnight operation. The properties of the organic membranes were not significantly different
from those of the inorganic, except that the inorganic had higher fluxes and rejections of color
and TOC. These results did not indicate any obvious advantage of the Kelco or the Pluronic
polymers over the hydrous oxides.
In comparing reverse osmosis treatment of simulated decker filtrate with ultrafiltration of
bleach plant effluent (E^ using hydrous Zr(IV) Si(IV) oxides, it was observed that fouling
seemed to be less of a problem in the ultrafiltration trials. The simulated decker filtrate usually
contained three to ten times as much of the potentially fouling constituents as the Ej, if TOC and
total solids are any indication. Thus, it is possible that the difference in fouling was caused by the
differences in feeds, rather than membrane properties. An indication that this is at least partly the
cause is seen in Figure 20, in which an ultrafiltration trial with both feeds is summarized. The flux
35
-------
recovered when the feed was changed from simulated decker filtrate to bleach plant effluent. This
comparison was partially biased because the simulated decker filtrate had been neutralized prior to
the test, and it was noted in later work that reactions of the feed constituents with the sulfuric acid
used for pH adjustment can produce species which adversely affect membrane performance.
TABLE 3. ULTRAFILTRATION OF BLEACH-PLANT EFFLUENTS BY MEMBRANES,
DYNAMICALLY FORMED OF KELCO BIOPOLYMERS
(0.27 p Selas ceramic tubes)
Hydrous*
Zr(IV)-SMlV)
Kelgin MV Kelgin XL Kelzan Oxide
Hyperfiltration properties
(0.05 M NaCI, 950 psi, 15 ft/sec, 25°C, pH ~tt
Flux, gpd/ft2 200 100 45
R _u,, %, conductivity 47 67 75
Ultrafiltration of bleach-plant effluent
(200 psi, ~6-ft/sec, 55°C, pH ~9)
Flux, gpd/ft2
Filtrate Analysis
TOC, ppm
Hobs' %
Color units
R Qbs, %
Chloride, molarity
Hobs' %
84
48
92
100
98
0.012
22
86
46
92
125
97.5
0.0093
39
52
41
92
110
98
0.0088
43
80
28
95
8
99.8
0.012
20
*19-tube bundle. Other test sections single tubes.
-------
IUU
90
~
~ 80
v>
o
70
60
^ 60
%
1*0
X
3 20
u
1 .
-
-
-
-
"
c
-
_
PULP WASH EFFLUENT
( pH 7.3 )
1
A
i 1 . 1 , 1 , 1 . 1 .
'
»
.
, ll , 1 , 1 , 1 , 1 ,
4
u
o "
BLEACH PLANT EFFLUENT -
( pH 8.5 )
TOC o
COLOR °
* SULFUR
I
4
1
i I . 1 , I , 1 i 1
0 ° _
> _
i I i | i I i 1 . 1
10
20
30 40
50 60
HOURS
TO 80 90 100
- 2.0
- 1.0
110
Figure 20. Comparison of ultrafiltration and kraft pulping of effluents.
Hydrous Zr(IV) oxide membrane on 19-tube Selas ceramic bundle.
( 5.6 ft/sec ; 200 psi ; 60°C ; ! Water Rinse )
(1.7m/sec;1.38x106Pa) '
-------
Summary
The laboratory tests resulted in several conclusions that were applied to the design of the
pilot plants.
1. The most desirable combination of processes and waste waters to be treated is reverse
osmosis treatment of decker filtrates and ultrafiltration of bleach plant caustic
extraction stage effluent (Ej).
2. The best dynamic membrane for reverse osmosis known at this time is a dual layer of
hydrous zirconium(IV) oxide and polyacrylic acid (Zr[IV] PAA).
3. The best dynamic membrane for ultrafiltration appears to be a dual layer of hydrous
zirconium (IV) and silicon (IV) oxides (Zr[IV] - Si[IV]).
4. The most favorable membrane support at the time the pilot plant was designed was the
Selas ceramic tube. The Mott stainless steel tubes then available were too unstable and
required filteraids. Steel tubes developed later had better characteristics.
5. In extensive tests using various circulation velocities and comparing the resulting fouling,
concentration polarization, pressure drops, and fluxes, a velocity of 4.57 m/s
(15 ft/sec) was a good compromise.
6. No adjustment of pH is necessary as long as the pH does not stay above 10 while using
the Zr(IV) PAA membrane. High pH is less critical to the Zr(IV) Si(IV) membrane.
38
-------
PI LOT PLANT STUDIES
The reverse osmosis and ultrafiltration pilot units were constructed at Oak Ridge National
Laboratory and shipped to International Paper Company's Mobile, Alabama mill. The units were
subsequently moved to the Company's Moss Point, Mississippi mill, which produces a variety of
bleached hardwood and softwood products. The units were located inside the pulp mill screen
room, with all power, instrument air, fresh and waste water feeds supplied to this location. The
reverse osmosis unit was installed to treat combined excess filtrate from the softwood and hard-
wood decker units, which are the last pulp washing and thickening units before the pulp is trans-
ported to the bleaching sequence. The ultrafiltration unit was installed to treat the excess wash
water from the caustic extraction stage in the bleach plant sequence and was piped to receive flow
from both the hardwood and softwood lines to ensure flow if one bleach line was not operating.
Equipment Descriptions
Reverse Osmosis Unit
The reverse osmosis unit is a modified version of a desalination pilot plant used by Oak Ridge
National Laboratory for tests performed for the Office of Saline Water at the brackish water facility
in Roswell, New Mexico (see Figures 21 and 22). The system consists of a 1.14 m3 (300 gallon)
stainless steel storage tank, a Gaso positive displacement pump (model 3466), and a stainless steel
piping loop arranged to hold 32 Selas 19-tube and 12 Selas 61-tube bundles (only the former were
used). The storage tank is equipped with drains, an overflow outlet, recycle lines for both concen-
trate and permeate, and water level controls to turn off the pump at excessively low levels. The
operating level is maintained by controlling raw waste water (decker filtrate) feed to the tank. The
pump is capable of about 7.89x10"3 m3/s (125 gpm) at 2.07x107 Pa (3000 psig) using a 5.6x104
watt (75 hp) motor, although the actual operating pressure was limited to 6.894x106 Pa (1000 psig).
The pump was equipped with a Greer Pulse-Tone surge dampener to smooth out pumping pulsa-
tions. The piping is arranged in a continuous loop, with the valving configuration designed to
arrange the tube bundles into groups of eight. Each group is comprised of two parallel series of four
bundles (see Figures 23 and 24). When fully loaded with 32 bundles, the surface area available for
membrane formation is about 3.72 m2 (40 sq ft). A single pass heat exchanger was used to hold
the temperature constant. Also, low tank level, high temperature, high and low pressure, and
safety pump cutoffs were installed to protect the pump and other system components. The con-
centrate flow from the pump can be controlled to produce the desired flow velocity (4.57 m/s
[15 ft/m]) across the tubes, and may be either discarded or recycled to the storage tank. The
permeate could be individually or collectively recycled or discarded, or any individual bundle could
be isolated. The mode of operation used here was to recycle the concentrate and discard the
filtrate.
Ultrafiltration Unit
The ultrafiltration unit (Figure 25) consists of three sections: a 0.76 m3 (200 gallon) Nalgene
storage tank, the pump, and the piping which held the tube bundles. The storage tank is equipped
with drains, low and operating level controls, overflow line, and feed and recirculation lines. The
feed line supplies fresh caustic extraction stage (Ej) waste water from the bleach plant, and the
recirculation lines allow recycle of either/or both permeate and concentrate to the tank. The op-
erational method was to recirculate the concentrate and discard the permeate. The pump was a
2.238x104 watt (30 hp) Goulds model 3933 multistage centrifugal pump, equipped with Noryl 2
(fiberglass reinforced plastic) impellers, and capable of flows of up to 0.19 m3/min (50 gpm) at
39
-------
BYPASS
CONTROL
VALVE
CONCENTRATE
RECYCLE
FEED SOLENOID VALVE
+4-
^FUOAT OPERATED
-* ;PAPER MILL WASTE STREAM
FEED
TANK
FILTRATE RECYCLE
. FEED VALVE
FLOAT CONTROL
n U/ LOW LEVEL
*- SL FLOAT CUTOUT
FILTRATE
COLLECT
MEMBRANE MODULES
RACK ASSEMBLY
, , Hrf VIBRATION PADS
LOOP DRAIN
«*«»-
FILTRATE
FLOWMETER
FILTRATE
MANIFOLD
( EACH MODULE BYPASSED )
Figure 21. ORNL-IP Co. hyperfiltration loop schematic.
-------
Figure 22. Pilot plant installation. Foregroundreverse osmosis (hyperfiltration) unit;
backgroundultrafiltration unit.
-------
FILTRATE
PROCESSt
INLET I
JACKET SIZE
IPS STAINLESS STEEL
CONSTRUCTION
2 19-TUBE BUNDLES
PER MODULE
FILTRATE
FILTRATE*
t PROCESS
' OUTLET
Figure 23. Hyperfiltration module assembly containing eight 19-tube bundles:
four assemblies in rack.
-------
Figure 24. Close-up of reverse osmosis unit.
-------
CONCENTRATE RECYCLE
FEED SOLENOID WASTE
Jtxt
BYPASS
FEED VALVE
FLOAT
CONTROL
PRES
CONTROL/
BYPASS
DRAIN
VALVE-FLOAT
OPERATED
LOW LEVEL
FLOAT
CUTOUT
PUMP
START
AND
CUTOUT
RELAY
STREAM
HMULTISTAGE CENTR. PUMP
CITY
W/VTERi
RECYCLE
4^ ROTA
DISCH
MANUAL
CONTROL
t*l
VALVE
-IHEAT EXCHANGER }-
+
MEMBRANE MODULES
RACK ASSEMBLY
INLET
MANUAL
VALVE
M
30 Hp
MOTOR
FILTRATE
RECYCLE
DRAIN
FILTRATE
>
MANIFOLD
FILTRATE
COLLECT
FILTRATE
ROTA
(EACH MODULE BYPASSED)
Figure 25. IP ultrafiltration loop schematic.
-------
pressures up to 5.52x106 Pa (800 psig). The pump was protected by cutoff controls for high and
low operating pressure, high temperature, high and low flow, and low tank level. The waste water
was pumped through a single pass heat exchanger for temperature control and into the piping array.
The piping was designed to hold sixteen 19-tube bundles in series in groups of four, in a configura-
tion similar to that used in the reverse osmosis unit. The valving allows bypass of any group of
four bundles. In addition to the Selas tube bundles, provision was made to accommodate a Union
Carbide UCARSEP module and an experimental module using Union Carbide carbon tubes.
Operating Experiences
The two units were completely separate but had several common operational characteristics.
The two pilot units employed tube bundles for membrane supports comprised of 19 Selas ceramic
(0.27 ym) tubes. The tubes were 0.53 cm (0.21 in) O.D. and 33 to 38 cm (13 to 15 in) long, with
one end open and embedded in epoxy and the other plugged with polypropylene plugs and fastened
into a spacing web to keep tube surfaces separated (see Figure 26). The membrane was formed on
the outside, and the permeate passed through the tubes into the fitting in which the bundle was
embedded. The total surface area for a complete bundle was 0.11 to 0.13 m2 (1.2 to 1.38 sq ft),
respective to the varying tube length.
The units were connected to a common power source in the pulp mill, and the frequent
power outages that occurred shut down both units. The pilot plants did not have provisions for
automatically restarting, so they would remain idle until someone arrived to restart them. Thus,
an intermittent power stoppage could shut down both pilot plants for a period of a day or a
weekend.
The pilot units were provided with systems to control pH, using sulfuric acid for the reverse
osmosis unit and acidic chlorination stage bleach plant effluent for the ultrafiltration unit. These
were provided to protect the stability of the membranes, especially the hydrous Zr(IV) oxide-
polyacrylate, and to minimize deterioration of the ceramic tubes. However, neither system was
used because the pH of both waste waters tended to stay below 10, except for brief increases to
approximately 11 with the bleach plant effluent. The tube formulation was adjusted by Selas to
make the ceramic tubes compatible with the hot alkaline solutions.
The waste waters fed to both units contained much higher concentrations of pulp fibers than
anticipated, compared with the laboratory simulated waste waters. The Oak Ridge National
Laboratory personnel concluded that no prefiltration was necessary, based on their experience
with the laboratory tests. The waste water used for most of the laboratory studies was diluted
weak black liquor which had been shipped by truck in a 55-gallon drum. The actual fiber load in
the process waste water was never seen, as the solids had settled in transit. The decker filtrate would
tend to have more fibers in it because the washing process involves vacuum dewatering on a rotat-
ing screen and because pulp spills in the screen room are washed into the decker filtrate sumps
during cleanup. These fibers tended to collect on the spacing web at the end of the bundle, block-
ing off the openings. The flow path was then progressively altered as fibers accumulated within
the interstitial spaces inside the bundle, filling the spaces and covering part of the membrane surface
(see Figure 27). In addition, the trapped fibers and reduced flows within the bundles were con-
ducive to biological growth, especially during periods when the unit was shut down. Microscopic
examination of the slime found in a tube bundle showed evidence of bacterial and flagellated
45
-------
0>
Figure 26. Selas ceramic tube bundles.
Top-used bundle with fiber accumulation; bottom-new bundle with several tubas broken ott.
-------
Figure 27. Broken tube bundle with moderate fiber accumulation.
-------
protozoan populations. These microorganisms have apparently adapted to conditions within the
pulp mill and were able to survive the high temperature, pressure, velocities, and pH conditions
encountered in the tube bundles, once the fiber build up provided them with a substrate.
A critical operational problem of both units was the tube breakage encountered when loading
and operating. The tube size, arrangement in the bundle, and the piping size (into which the
bundles were loaded) were coordinated to provide as even a flow throughout the bundle as possible.
The outer ring of tubes were very close to the inside of the piping in order to force the flow into
the interior of the bundle. The bundles were cemented into machined-out threaded reducing plugs
and screwed into tees on the ends of the module piping (see Figures 26 and 28). Many bundles
were broken and had to be patched or replaced because of outer ring tube breakage during the
screwing of these plugs into the tees. Often, bundles were broken during membrane formation and
operation and had to be removed or simply isolated. A broken bundle can be critical during
membrane formation because the flow of clear concentrate through a broken tube into the
common filtrate manifold will falsely indicate a high flux rate and cause the operator to allow
a thicker than desired membrane to form. This is important because the overall flux rate is a
parameter in judging when the hydrous Zr(l V) oxide layer formation is complete. A broken bundle
during actual operation is more easily detected because the permeate flow will increase and the dark
concentrate will color the clear permeate.
Unfortunately, a broken bundle cannot simply be isolated and forgotten, especially with the
high pressure reverse osmosis unit, because broken tubes tend to break away from their bundles.
They are then forced through the piping from bundle-to-bundle breaking more tubes until all the
bundles are broken, or bits of tubes are trapped in the piping, completely blocking the flow and
forcing a shutdown of the unit.
In addition to the breakage problems, some bundles showed signs of poor membrane forma-
tion because tubes within the bundle touched each other and/or the inside of the piping. This
phenomenon could also cause false flux and rejection values because membrane imperfections at
these places would allow excess flow and poor rejections, even though a good membrane was
applied to the rest of the tube. Thus, the tube alignment, size, and bundle-to-piping tolerances are
critical to successful membrane formation.
The tube breakage, which caused a number of delays in the pilot plant work, was at first
attributed to weak tubes. It was theorized that vibration and pumping pulsations, coupled with
the hot alkaline solutions, caused them to break. Also, because most tubes broke at the threaded
fitting end, it was also believed that the piping arrangement, which introduced flow at right angles
to the bundles at this point, was contributing to excess breakage. The manufacturer performed
extensive testing to check these theories and determine whether the ceramic formulation needed
adjustment. In these studies the critical harmonic frequencies of tubes were calculated, and ex-
tended tests of tube lengths necessary to produce these sympathetic vibrations were performed.
Repeated starting and stopping tests were also conducted to determine the effect of high pressure
surges on unprotected tubes. In these, as well as in accumulative fatigue tests, no tubes were
broken. Bubble point tests of tubes, before and after bundle fabrication, showed that the ceramic
formula was within specification. Thus, the ceramic material itself was not at fault.
The breakage of tubes upon loading was traced in part to the piping into which the bundles
were inserted. The threaded joints between the module pipes and the tees leaked at high pressure
48
-------
CO
Figure 28. Tube bundle installation: reverse osmosis unit.
-------
and were reinforced with additional welding. This caused a slight swelling of the metal inside the
pipe at the point of the weld and also caused the alignment of the tee and the pipe to shift off-
center slightly. This problem, plus the handling during shipping and installation, altered the
tolerances to the point that the outer tubes rubbed the inside of the piping during installation.
This contributed to both breakage and membrane attrition. Imperfections in the bundles also aggra-
vated the problems created by the piping. During the pilot work, it was found that the die for tube
formation had worn, causing slightly oversized (by about 0.030 inch) diameter tubes to be made.
These bundles could be installed only after several of the outer tubes were removed. Another ship-
ment of bundles contained some which were not aligned perfectly and could not be screwed into
the piping. The bundles were used by altering the screw-in plug to a clamp-in arrangement.
It should be noted that these problems have been eliminated in newer systems built by Selas
Corporation. The actual configuration is proprietary, but it consists of ceramic tubes which are
clamped into a module with beveled seals, removing the necessity of screwing bundles into the
module. The concentrate flow is still outside the tubes and with adequate prefiltration, modules
of this type have been operated successfully in a pilot installation for over 9000 hours. Thus, the
ceramic tube is a viable dynamic membrane support and should not be underrated because of
mechanical difficulties experienced while operating a pilot plant designed in the beginning of the
development of such systems.
MEMBRANE FORMATION PROCEDURES
Cleaning of Units
The formation of dynamic membranes requires that no contaminants be present in the
solution from which the membrane is laid. Extensive preformation cleaning of the wetted surfaces
is necessary prior to the membrane formation step. The usual washing procedure for both units
involves circulation of basic and acidic solutions at operating pressures and flows through the entire
system. The solutions used were 0.5 JW NaOH (circulated for several hours) followed by rinses with
mill supply water, and 0.5 M HNO3 (circulated for several hours) followed by repeated rinses with
deionized water, until the conductivity of the rinse solution approached that of fresh deionized
water. When new tube bundles were installed for a run, this cleaning procedure was performed
prior to their installation. When the wash was intended to prepare existing tubes for a new mem-
brane, or was simply intended to improve flux rates, the tubes were washed in place. When a wash
was intended for flux improvement only, a solution of Na2C03 has been found to be beneficial,
without the problems associated with extremely basic or acidic solutions. The importance of clean-
ing the units before installing new tube bundles, versus cleaning after installation, was reinforced by
the results in Run U-5. In this case, the material being washed off of the piping was apparently
deposited on the tubes, adversely affecting the performance of the initial membrane.
Forming the Membranes ,
Both units use a dual layer membrane that is dynamically formed on the porous ceramic
tubes by circulating a solution containing the membrane-forming material over them under pressure
and at a surface velocity sufficient to minimize concentration polarization. Both units use the same
50
-------
sublayer material (hydrous zirconium oxide), although it is applied at the pressure peculiar to each
unit's operation. The parameters used to determine the progress of the membrane formation are
conductivity rejection (% R) as defined earlier and the membrane flux. The flux of new ceramic
tubes with no membrane is quite high compared with that of a fully formed membrane, and the
conductivity rejection should be zero. By monitoring the rise in % R and the drop in flux and
comparing these values to those obtained in laboratory experiments, it is possible to control the
formation step. In the pilot plant work it was observed that higher concentrations of the membrane
materials in the formation solutions were necessary to form "good" membranes, especially when
reforming old membranes, than was predicted from laboratory scale tests. The cause for this is not
clear because pilot work with desalination units did not show such aberrations. However, adequate
membranes were formed by adding more of these materials (in some cases over eight times the
predicted amounts) to the formation solution. This phenomenon was more noticeable with the
zirconium oxide layer formation than with either of the secondary layers. Unfortunately, time
did not allow a complete study.
Procedure: Reverse Osmosis Unit
The unit is washed and filled with deionized water. Sufficient NaNO3 is added to raise the
conductivity of the solution to about 5000 ymho/cm (concentration about 0.05 M NaNO3), and
the pH is adjusted to 3.9 to 4.2, using NaOH or HNO3. The pH is critical because this is below the
isoelectric point of hydrous zirconium oxide, and the colloidal suspension is least stable in solution
at this pH. Hydrous oxide membranes have been found to perform best when formed under these
conditions. The unit is started up and this solution is circulated at the normal operating conditions
of6.894x106 Pa (1000 psig) and 4.56 m/s (15 ft/sec). The zirconium is added as Zr(N03)4 in
sufficient quantity to achieve a concentration of about 10~4 M. The pH usually must be readjusted
upward when the Zr(N03 )4 is added. This solution is circulated over the tubular supports, with all
concentrate and permeate flow being recycled, until the flux drops from its initial high level
(usually 8.15 m3/m2/d [2000 gfd]) and the salt rejection (measured as conductivity) begins to rise.
Ideally, a completed hydrous Zr(IV) oxide membrane should have a conductivity rejection of about
30 to 40% with fluxes of about 1.22 to 2.04 m3/m2/d (300-500 gfd). In actual practice, the flux
and conductivity rejections occupy much wider ranges. If the membrane performance is not satis-
factory, longer formation times or extra Zr(N03 )4 may be used.
Once the hydrous Zr(IV) oxide membrane is complete, the solution is drained out, and the
system is rinsed with deionized water. A fresh NaNO3 solution is then prepared, but the pH is now
adjusted to 2.0 to 2.3 using HIM03. The unit is again started and operated at the usual pressure and
flow. The polyacrylic acid (Rohm and Haas Acrysol A3, molecular weight about 150,000) is
added to the tank (about 150 to 180 grams/0.95 m3 of water) and the unit allowed to run until the
flux drops to approximately 5.09 m3/m2/d (125 gfd). The pH is then slowly raised in increments
of about 0.5 to 1.0 units, allowing the parameters of flux and conductivity rejection to come to a
steady state after each adjustment. If the flux does not drop, or if the conductivity rejection is too
low, extra polyacrylic acid may be used. In extreme cases, the membrane can be washed off and
another formed. Successfully completed membranes should have fluxes of 2.04 to 4.07 m3 /m2 /d
(50 to 100 gfd) with conductivity rejections of 88 to 92%. The unit is then rinsed prior to use.
Procedure: Ultrafiltration Unit
The system is washed and filled with deionized water. As with the reverse osmosis unit,
NaN03 is added to provide reference conductivity and the pH is adjusted to 3.9 to 4.2 prior to the
51
-------
addition of Zr(N03)4 (to obtain a 10~4 M solution). The membrane formation is performed at the
same velocity, 4.57 m/s (15 ft/sec), but with a higher pressure than that of the usual ultrafiltration
operation. The membrane is formed at 2.76x106 to 3.10x106 Pa (400 to 450 psig), versus the
normal operating pressure of 1.59x106 Pa (230 psig). The membrane formation step is continued
until the flux drops to 8.15 to 16.3 m3/m2/d (200 - 400 gfd) and the conductivity rejection
increases to about 20 to 30%. If performance is not satisfactory, longer formation times or extra
Zr(NO3)4 can be used. After completion of the hydrous Zr(IV) oxide layer, the solution is
drained and the unit rinsed with deionized water. Then another NaN03 solution is prepared, the
unit is started again at an elevated pressure, the pH is adjusted to about 8, and approximately
350 g/0.57 m3 H2O of sodium silicate solution (Na2Si03, Matheson, Coleman Bell, 40 - 42° Be)
is added. The unit is operated until the flux drops to about 4.07 m3/m2/d (100 gfd) and the
conductivity rejection increases to 30 to 40%. If these conditions are not reached, additional
formation time or extra Na2SiO3 may be added. The system is then rinsed prior to use.
52
-------
SECTION 6
RESULTS
REVERSE OSMOSIS UNIT
The operation of this unit was erratic due to power failures, mechanical problems, and tube
bundle breakage. The results given in the following tables and figures must be interpreted in con-
junction with the operating times, shutdowns, and subsequent lags in operation. The results of the
first successful trial. Run H-1, are shown in Figure 29 and Table 4. This run was limited to eight
tube bundles and the membrane formation was abnormal. Excess Zr(N03)4 and polyacrylic acid
were necessary, and the final fluxes and percent rejection were lower than desired at the end of the
formation step. Two bundles showed signs of breakage (high flux and low percent rejection com-
pared with the others) but later approached the same operating levels, as if the membrane had
repaired itself. This occurred a number of times during the project and was possibly caused by tube
bundle imperfections. The hypothesis is that tube imperfections act as oversized pores that will fill
in and become covered with membrane more slowly than the usual tube surfaces.
The unit was started with decker filtrate using no pH adjustment (pH = 10.3) and operated at
a pressure of 6.894x106 Pa (1000 psig) and a surface velocity of 4.57 m/s (15 ft/sec). The initial
flux was 2.44 m3/m2/d (60tO gfd) with a conductivity rejection of 77%. After 12 hours of opera-
tion these values had dropped only slightly, and problems with system controls shut down the unit.
Because the unit was on and off several times over a span of a few days, the feed tank was drained,
refilled with fresh decker filtrate, and restarted on February 9, 1977 (the beginning of Run H-1).
Shortly after startup, two of the eight bundles broke and had to be isolated. Excessive pump
leakage and power failures stalled the restarting until the next day. Over the next 4 days of
operation, the flux dropped from 2.35 to 1.31 m3/m2/d (57.6 to 32.2 gfd). The unit was again
shut down for a pump repair and restarted on February 15,1977 with the flux on startup about
equal to that before the shutdown. The unit ran for 2 more days until the seals on the feed pump
pistons began leaking excessively. This did not cause the unit to shut down, but as concentrate
was lost from the pump, raw feed was introduced at a rate higher than permeate production caus-
ing undue dilution of the concentrate. The data indicate the rate of concentration of the feed
constituents was adversely affected. The run was terminated when the other tube bundles began
to break. The bundles were removed and found to contain fiber accumulations.
The reverse osmosis unit was next operated in April, when replacement tube bundles became
available. In Run H-2, 29 bundles were loaded, but several were broken during membrane forma-
tion and in an initial trial. The broken bundles were removed and found to contain fiber accumula-
tions after only 5 hours of operation. Also, several tubes showed signs of rubbing against the inside of
the piping. The unit was started up on May 2,1977, with a 50 ym rating filter bag placed on the
raw feed line. Because the unit had been left dry during the 2-week downtime, the membrane
53
-------
75
K 50
S?
25
100
75
25
100
75
25
100
75
5»
25
60
a
X 30
1.
100
75
* 50
25
12
9
a. 6
3
19 a id a
D COLOR
TOTAL SOLIDS
Q D u u
D ASH
u u u a
P TOC
a a
- DO
CHLORIDES
* 9
FLUX ~
_. * * * -
~ CONDUCTIVITY
^
_ . *
PH
I 1 1 1 1 1 1 1 1 1 1
2.0
1.5 5
1.0 E
0.5
20 40 60 SO 100 120 140
TIME, hours
160
180
200
220
240
Figure 29. Run H-1. Reverse osmosis of decker filtrate.
6.894 x 10 Pa,
1000 ptig
4.57 m/s
15 ft/sec
52bC.
pH 8-10
54
-------
TABLE 4. REVERSE OSMOSIS TREATMENT OF DECKER FILTRATE, RUN H-1
01
01
Date,
1977
2-9
10
11
14
15 '
16
17
18
21
Elapsed
Operating
Time, hr.
1
51
22
95
1012
118
142
1663
238
Flux,
m3/m2/d
(gfd)
2.44 (60)
2.35 (57.6)
1.98(48.7)
1.31 (32.2)
1.27(31.2)
1.19(29.1)
0.92 (22.7)
0.81 (19.8)
PH
Concen-
trate
10.0
9.7
9.8
9.4
10.4
8.7
8.4
8.4
8.3
Per-
meate
10.4
10.2
9.3
9.8
10.2
9.6
8.5
9.2
9.5
Conductivity -
Concen-
trate
530
680
920
370
800
980
1200
1300
1400
Per-
meate
100.0
110.0
110.0
57.0
140.0
70.0
56.0
70.0
88.0
~~~" Total Solids, p.p.m. Ash, p.p.m.
% R Concen- Per- % R Concen- Per-
trate meate trate meate
81.0 Startup
83.8 Restart
88.0 457 10.0 97.8 257 9
86.2
82.5
92.8 1444 19.0 98.7 680 17
95.3 1230 4.0 99.6 423 9
94.6 1647 12.0 99.3 572 6
93.7 - - -
_. _
%R
96.5
97.5
99.1
99.0
-
(continued)
Date,
1977
2-9
10
11
14
15
16
17
18
21
Elapsed
Operating
Time, hr.
1
51
22
95
1012
118
142
1663
238
Volatile Solids, p.
Concen-
trate
200
764
807
1075
-
Per-
meate
1
2
0
6
-
p.m.
%R
99.5
99.7
100.0
99.4
-
Concen-
trate
2440
3720
3160
4000
-
Color, c.u
Per-
meate
0
0
2.5
0
-
% R Concen- Per- % R
trate meate
100.0 42 3 92.9
100.0 28 2 92.9
99.9 24 5 79.2
100.0 24 5 79.2
- - - -
'Unit down for power failure and repair. Restarted 2/10/77.
Unit down to repair pump. Restarted 2/15/77.
Found pump piston seals leaking excessively. Concentrate in tank diluted by excess raw feed.
-------
performance may have been altered. The initial flux was 3.52 m3/m2/d (86.4 gfd) and the flux on
startup on May 2, 1977 was 3.30 m3/m2/d (81.0 gfd). The flux dropped to 1.53 m3/m2/d (37.6
gfd) within 18 hours and declined to 0.46 m3/m2/d (11.3 gfd) within 97 hours. At this time, the
power to the unit was shut off and did not become available again until May 10,1977. Mechanical
problems delayed startup until May 11,1977, and the membrane initially showed no significant
difference in performance as a result of the delay. The next day the tube bundles started breaking,
the flux was low, and extreme pressure fluctuations occurred. The unit was shut down and rinsed
the next day which resulted in more tube breakage. A subsequent wash with a Na2C03 solution
(pH = 10.6) produced a slight improvement in flux and the loss of more tube bundles. An attempt
at restarting with fresh decker filtrate resulted in a flux of 1.80 m3/m2/d (44.1 gfd), but a power
failure and the following restart broke all but three of the remaining tube bundles. On shutting
down the unit and removing the broken tubes, it was discovered that fibers had again collected in
the tube bundles and that biological activity was producing a slime layer and swelling up the fiber
mats, producing a stress on the tubes. It was later found that the fibers had gotten into the system
when the filter bag clogged up during a high fiber load period and that raw decker filtrate had over-
flowed into the tank. The data for Run H-2 are summarized in Figure 30 and Table 5.
During a wash sequence to prepare the unit for another trial, the surge dampener on the pump
outlet sprang a leak. On disassembling the unit, the dampener was found to be irreparably damaged
by corrosion. A replacement surge dampener was ordered from Eaton Corporation and was
installed in early September. The unit was loaded with tube bundles and a NaOH wash was performed.
The hydrous Zirconium oxide membrane was formed using an excess of Zr(IM03)4, and this
resulted in a lower than desirable flux and percentage of rejection. There were several power
failures during the polyacrylic acid (PAA) membrane formation. The solution being circulated
formed a black precipitate as the NaOH was added to raise the pH, indicating that iron was probably
being dissolved from the surge dampener. The flux was initially 5.5 m3/m2/d (135.3 gfd) before
the PAA was added, rose during the formation step, and dropped to a final level of 3.39 m'/mVd
(83.2 gfd) with a conductivity rejection of only 8.75% (at pH = 8.65). It appeared that the
dissolved iron had coagulated the PAA, because the solution contained clumps of brownish-black
material which were not present in previous trials.
This membrane's low rejection at a fairly high flux indicated a poor formation, so the unit was
washed out with a NaOH solution (pH = 12). Then another PAA formation step was attempted,
with emphasis on minimizing the exposure of the carbon steel surge dampener to the low pH
(pH = 2) solution. This formation step followed the same pattern as the preceding one, with a final
flux of 5.04 m3/m2/d (123.6 gfd) and a conductivity rejection of only 20.8%, again indicating a
poor membrane. The surge dampener was removed, cleaned, and coated (wetted parts) with an
epoxy-base paint to prevent corrosion. The surge dampener was reinstalled and several of the
broken tube bundles removed. The signs of wear on the sides of the bundles indicated that the
tubes had been rubbing the piping, contributing to poor membrane formation.
The formation of the PAA membrane was completed, with a final flux of 4.38 m3/m2/d
(107.6 gfd) and a low conductivity rejection of 28%, indicating that the membrane was still not optimal.
It was decided to try to use this membrane, because the rubbing tubes might have made it impossible
to form a better one. Run H-3, summarized in Figure 31 and Table 6, began on October 28,1977
using fresh decker filtrate filtered through a 10 ym rating filter bag. The unit achieved a flux of
4.12 m3/m2/d (101.1 gfd) on startup, giving a brown initial filtrate which cleared up in about 15
minutes. The conductivity rejection was then 89.3%, at a flux of 3.72 m3/m2/d (91.3 gfd). The
56
-------
IUU
75
*50
25
100
75
= 50
as
25
100
75
as
25
100
75
c 50
a?
25
100
I75
X 50
E! 25
100
75
-50
25
12
9
a 6
3
D U
-
a
-
u
-
D
-
-
~
-
0
,
-
~
. '
-
! 1 I 1
COLOR
TOTAL ORGANICS
ASH
TOTAL SOLIDS
FLUX
CONDUCTIVITY
pH
D
D D
D °
a a
~
.
e «
I
4.0
2.0 e
1.0
20
40
60
80 100 -HNo operation due to >
TIME, hours power loss)
220 240
Figure 30. Run H-2. Reverse osmosis treatment of decker filtrate.
6.894 x 106 Pa 4.57 m/s 52°C.
1000 psig 15 ft/sec pH 8-10
(ACTUAL ELAPSED OPERATING TIME 126 HOURS.)
57
-------
TABLE 5. REVERSE OSMOSIS TREATMENT OF DECKER FILTRATE, RUN H-2
Date,
1977
5-2
3
4
6
11
12
Elapsed
Operating
Time, hr.
1
18
47
971
103
126
Flux,
m3/m2/d
(gfd)
3.30(81.0)
1.53 (37.6)
0.82 (20.2)
0.46(11.3)
0.51 (12.5)
0.37 (9.1)
pH
Concen-
trate
9.9
9.9
10.1
9.7
7.7
8.1
Per- Concen-
meate trate
9.9 1 100
9.9 2200
9.7 3200
9.4 3200
7.4 2600
7.8 2700
nductivity
Per-
meate
130
260
420
380
600
490
_ .A ...
% R Concen- Per-
trate meate
88.2 Startup
88.0
86.9 5579 278
88.1 5938
76.9 4615 408
81.9 4758 387
m.
%R
95.0
-
91.2
91.9
(continued)
Date,
1977
5-2
3
4
6
11
*"2.._, ..
Elapsed
Operating
Time, hr.
1
18
47
971
103
126
Ash
Concen-
, p.p.m.
Per- % R
trate meate
Startup
1810
1961
1527
1815
64 96.5
170 89.9
1 1 1 93.9
Volatile Solids, p.p.m.
Concen- Per-
trate meate
3769 214
3977
3088 238
2943 276
%R
94.3
92.3
90.6
Color, c.u.
Concen- Per- % R
trate meate
19,600 45 99.8
18,400 45 99.8
19,600 250 98.7
21,800 480 97.8
Shut down due to power failure.
-------
75
25
75
25
75
25
1UM
75
25
75
25
9
i 6
3
3 90
<-60
L 30
O
- o
v o a
V O
I
h *
1
s
I
ll
i
a a a o
* »
g a U B
0 0
OTOC
a SODIUM
|g
1
1
1
\t «
,_,,. 1
0 COLOR j
a VOLATILE SOLIDS ,
CHLORIDES
D COD
O TOTAL SOLIDS
o o
\
\
1
1
1
1*
OBOD;
D TOTAL SULFUR
ASH
1
1
1
CONDUCTIVITY
1
1
U
1
1
I.
\
PH
1
1
\
, ~~
1
FLUX
1
1 -
1 1 1
4.0
3.0-0
2.0 "e
1.0
170 180 200 220 240 260 | 0 20 40 60 80 100 120 140 160 180
« I TIME. HOURS
\Jnit down due to \Jn\t down to install
mechanical problems new tubes and membrane
Membrane washed
Figure 31. Run H-3. Reverse osmosis treatment of decker filtrate.
6.894 x 106 Pa 4.57 m/s 52°C
lOOOpsig 15 ft/sec pH8-11
(ACTUAL ELAPSED OPERATING TIME BEFORE MEMBRANE REPLACEMENT 37 HOURS.)
59
-------
TABLE 6. REVERSE OSMOSIS TREATMENT OF DECKER FILTRATE, RUN H-3
Date,
1977
10/28
11/4
11/7
11/8
11/22
11/23
11/24
11/25
11/28
11/29
Elapsed
Operating
Time, hr.
JS
30
0.5 3
17
41
169
Flux,
m3/m2/d
(gfd)
3.72 (91.3)
3.18 (78.1)
3.42 (84.0)
5.05 (124.0)
3.95 (97.0)
2.73 (67.0)
1.86 (45.6)
1.40 (34.4)
1.74 (42.8)
1.09 (26.7)
pH Conductivity Total Solids, p.p.m.
Concen- Per- Concen- Per- % R Concen- Per- % R
trate meate trate meate trate meate
9.65
9.05
8.80
10.00
9.90
10.00
9.70
10.45
9.20
8.95 1400 150
9.00 1100 92
8.70 1400 200
10.20 1200 530
10.20 1500 500
10.40 2100 600
10.20 1600 600
10.80 1650 800
9.75 1150 500
89.3 1389 77
91.6 884 57
85.7
55.8 884 70
66.7 - -
71.4 2668 101
62.5 2648 96
51.5 2076 234
56.5 1415 '40
94.5
93.6
92.1
96.2
96.2
88.7
97.2
Ash, p.p.m.
Concen- Per- % R
trate meate
527 32
373 10
359 4
1 034 40
940 45
777 1 04
460 18
93.9
97.3
98.9
96.1
95.2
86.6
96.1
Chloride, p.
Concen- Per-
trate meate
24.6 3.4
25.3 4.1
30.7 3.5
37.2 3.1
55.9 4.9
59.1 6.4
53.6 8.3
29.7 3.3
p.m.
%R
86.2
83.8
88.6
91.7
91.2
89.2
84.5
88.9
(continued) '
Data,
1977
10/28
11/4
11/7
11/8
11/22
11/23
11/24
11/25
11/28
11/29
Elapsed
Operating
Time, hr.
Ji
30
0.53
17
41
67.,
1454
169
Flux,
m3/m2/d
(gfd)
3.72 (91.3)
3.18 (78.2)
3.42 (84.0)
5.05 (124.0)
.3.95 (97.0)
2.73 (67.0)
1.86 (45.6)
1.40 (34.4)
1.74 (42.8)
1.09 (26.7)
~- Color, c.u. ' Sodium,
Concen- Per- % R Concen-
trate meate trate
4575
1700
2230
6850
8050
7200
4250
27 99.4 177.0
15 99.1 108.0
0 100.0 143
- - 200.4
0 100.0 431.0
19 99.8 385.2
78 98.9 319.7
75 98.2 128.6
p.p.m., as Na 5-Day
Per- % R Concen
meate trate
0.5 99.7 430
10.0 90.7 210
2.6 98.2 220
11.1 94.5
25.1 94.2 440
21.2 94.5 440
54.7 82.9 480
3.3 97.4
BOD, p
- Per-
meate
66
58
58
87
64
45
.p.m. COD, p.p.m.
% R Concen- Per- % R
trate meate
84.7 1450
72.4 780
73.6 620
- 1389
80.2 2034
85.5 2679
90.6 2728
- 1468
575
96
69
104
135
129
161
88
60.3
87.7
88.9
92.5
93.4
95.2
94.1
94.0
(continued)
Date,
1977
10/28
11/4
11/7
11/8
11/22
11/23
11/24
11/25
11/28
11/29
Elapsed
Operating
Time, hr.
Ji
30,
0.53
17
41
1454
169
Flux.
m3/m2/d
(gfd)
3.72 (91.3)
3.18 (78.2)
3.42 (84.0)
5.05 (124.0)
3.95 (97.0)
2.73 (67.0)
1.86 (45.6)
1 .40 (34.4)
1.74(42.8)
1.09 (26.7)
TOC, p.p.m.
Concen- Per- % R
trate meate
54
340
480
334
495
807
812
666
474
40.3 25.4
30.5 91.0
53.7 88.8
27.9 91.6
30.0 93.9
47.0 94.2
40.0 95.1
47.4 92.9
29.2 93.8
Total Sulfur, p.p.m.
Concen- Per- % R
trate meate
42.5 1.6
41.3 1.7
67.0 3.3
33.0 1.0
42.0 1.0
87.0 1.3
83.0 8.2
70.0 2.4
42.5 1.0
96.2
95.9
95.1
97.0
97.6
98.6
90.1
96.6
97.6
Volatile Solids,
Concen- Per-
trate meate
862
511
525
1634
1608
1299
955
45
26
66
61
51
130
28
p.p.m.
%R
94.8
94.9
87.4
96.3
96.8
90.0
97.1
2Mechanlcal problems 10/31/77. Restarted with fresh feed 11/4/77.
Unlt shut down over the weekend.
restarted with new tubes and membrane.
Unit down. Cleaned membrane; restarted with fresh decker filtrate
-------
unit was found shut down on October 31,1977 because of a malfunctioning feed valve. The valve
was replaced, the unit flushed, and restarted with fresh decker filtrate on November 4, 1977. The
flux was 3.18 m3/m2/d (78.2 gfd) with a conductivity rejection of 91.6, and the unit was shut
down over the weekend, rather than risk an accidental shutdown.
The unit was restarted on November 7, 1977 with only 10 unbroken bundles left, obtaining
a flux of 3.42 m3/m2/d (84.0 gfd) and a conductivity rejection of 85.7%. The flux was higher the
next day, but more bundles were broken. The unit was shut down when a high pressure cutoff
turned off the pump. An attempt to restart it caused the safety rupture disk to blow out and shut
down the unit again. On removing the tube bundles, it was found that broken pieces of tubes had
jammed the piping and blocked the flow after only 30 hours of operation. Fibers were again caught
in the tube bundles, and a build up of pitch was noticed. The unit was cleaned out and the 9 avail-
able bundles installed. A new membrane was formed, and the unit restarted on November 22,1977.
The new membrane obtained a flux of 3.95 m3/m2/d (97.0 gfd) with a conductivity rejection
of 54.2%. The unit ran until the morning of November 28,1977, when it was found down with the
filter bag overflowing into the tank. The unit was drained and flushed with water and a NaOH
solution (pH = 11.9). The unit was restarted with fresh decker filtrate, achieving a flux of 1.74
m3/m2/d (42.8 gfd) and a conductivity rejection of 51.5%. The unit shut down again on
November 29,1977 because of several broken bundles. It was restarted with one bundle left
giving a flux of 1.09 m3/m2/d (26.7 gfd), with the pressure drop through the unit increasing,
indicating a progressive blockage. On November 30, 1977, it was found shut down with a blown
rupture disk. The reverse osmosis portion of this project was then terminated.
ULTRAFILTRATION UNIT
The operation of the ultrafiltration unit was also erratic because of power and equipment
failures. The data presented in the following graphs and tables must be interpreted along with the
fact that these repeated shutdowns prevented the unit from reaching any semblance of a steady
state. The unit first ran in September of 1976, but mechanical problems prevented the gathering of
much useful data. A Zr(IV)Si(IV) oxide membrane was formed, giving a flux of 2.93 m3/m2/d
(72 gfd) and a conductivity rejection of only 19%. This membrane was subsequently covered with
another complete dual-layer membrane in an attempt to improve upon performance. These par-
ameters did not improve, so the system was started on September 23,1976 with fresh caustic
extract (Et) using only nine bundles (the others had broken). The initial flux was 2.16 m3/m2/d
(26.6 gfd) in 2 days. On September 26, 1976, the pH control system malfunctioned, causing the
tank to overflow. The flux at this time was only 0.19 m3/m2/d (4.6 gfd), so the unit was rinsed
with water and a Na2 CO3 solution. During the rinse sequence September 28,1976, a pump bearing
failed and had to be replaced. The pump was repaired on October 12,1976, and caustic (1 JVI NaOH)
and acidic (1 M^HNO3) washes were performed. Then another membrane was applied, with 15
bundles in operation. During this formation step the pump was irreparably damaged, and the unit
was left full of deionized water until a replacement pump arrived.
On February 2,1977, the system was again operable, and fresh EI was added to the tank. No
pH control was used in this or subsequent runs. The unit was started up for a short time on
February 2,1977 and achieved a flux of 2.30 m3/m2/d (56.6 gfd). Over the next few days, the
unit was operated intermittently as control and other mechanical problems were corrected.
6T
-------
The flux gradually rose to 4.72 m3/m2/d (115.9 gfd) as the unit was operated. On
February 11,1977, the unit was filled with fresh Ex, and Run U-1 began. The flux dropped from
an initial value of 5.13 m3/m2/d (126 gfd) to 3.58 m3/m2/d (87.7 gfd) on February 14,1977.
The unit was shut down on February 15,1977 because of controller malfunction, and when it was
restarted it produced a flux of 3.28 mVmVd (80.5 gfd). The flux held fairly steady for the rest
of this run, in the range of 3.67 to 4.56 m3/m2/d (90 to 112 gfd), until the last few days of opera-
tion when the total solids content of the concentrate stream rose above 10% (see Figure 32 and
Table 7). The data from February 18,1977 to February 28,1977 are not representative of the
unit's capabilities because the unit was not operating for several days during this period, and at one
time the feed control valve jammed open, diluting the concentrate. The unit restarted on February
28,1977. The rejections of color, total solids, volatile solids, and ash were high throughout the
run. The flux dropped when solids content in the concentrate stream was increased (see Figure 33
and Table 8) as expected. However, the pressure drop through the unit did not increase signifi-
cantly when solids concentration was increased (see Figure 34 and Table 9).
The unit's operation after March 10,1977 was too irregular to generate useful data because
of numerous mill power failures. The tank was refilled with fresh waste water (Ex) and restarted on
March 15,1977, achieving a flux of 3.74 m3/m2/d (91.8 gfd) on March 16,1977 after approxi-
mately 12 hours of operation. The flux dropped to 3.10 m3/m2/d (71.1 gfd) 3 days later and a
water rinse was performed. The unit was restarted with fresh waste water (Ex) and was giving a flux
of 6.36 m3/m2/d (156.1 gfd) after 1.5 days. The operation of the unit was again irregular for the
next 5 days because of mechanical problems in the bleach plant, but each time the unit was
restarted the flux was greater than 4.48 m3/m2/d (110 gfd). Soon after this, tube bundles began
to break, and the unit was shut down to remove them. The bundles were found to be clogged with
pulp fibers. These fibers not only contributed to bundle breakage but drastically altered flow con-
ditions in and around the tubes. Unfortunately, no chemical data are available for the operation
after Run U-1. However, the effect of the water rinse on the flux is significant.
At this time no ceramic tube bundles were available, so the UCARSEP carbon tube module
supplied by the Union Carbide Corporation was installed. This module could not be used in con-
junction with the ceramic tubes because it had a pressure limit of 8.62 x 10s Pa (125 psig). The
unit was supplied with a proprietary membrane already applied, and it had been stored full of
water until this time. Prior to startup with the waste water (Et), the module was rinsed with mill
supply water. During this rinse, the rinse water stream turned brown for a short time, as if dark
suspended solids were being flushed out of the module. The unit was then started up with fresh
waste water (E}) at an inlet pressure of 7.93 x 10s Pa (115 psig). The permeate was initially a dark
brown and after an hour of operation, it was still as dark as the feed with a conductivity rejection
of 2.3%. It was concluded that the membrane material had loosened from the support during the
storage period and had sloughed off on startup. The UCARSEP module was not used further
because ceramic tube bundles were due to arrive shortly.
The newer tubes were fabricated into bundles by Oak Ridge National Laboratory personnel
and installed on April 25,1977. Then a membrane (Zr[IV]-Si[lll] oxide) was formed. Several
bundles were broken during the formation step. On removal it was found that the epoxy cement
used to join the bundles to the steel fitting had failed and allowed one entire bundle to slide into
another, damaging both. Another problem was that these tubes were slightly larger in diameter
than the previous ones, and the tubes which were removed showed signs of rubbing the piping.
This observation becomes more important when considered in conjunction with the fact that the
62
-------
100
75
C 50
ae
* 25
ifl/W\
100
75
CC 50
86 25
4/\/\
100
75
a? 50
25
mn
IW
75
a?
25
160
1,120
X 80
3 40
O) u-
f.« on
IAJ OU
60
* 4°
20
1*9
1*
9
a 6
3
° ° l
Q
L
a o 0 <
_
-
a a a u
-
-
-
_
*
_
<
-
-
_
_
1 1 1 1 1 1 1 1
D TOTAL SOLIDS
3 o a aoaao none
o ASH
0 TOTAL ORGANICS
(BY COMBUSTION)
o COLOR
CHLORIDE
* .FLUX . . ~
. *
CONDUCTIVITY
,.. - «...
pH
i i i i i i i i i i i i i i l l 1 i
6.0
4.05
E
2.0
20 40 60 80 100 120 140 160 300 320 340 360 380 400 420 440 460 480 500 520 540 560 580 600 620 640
TIME, hours
Figure 32. Run U-1. Ultrafiltration of caustic bleach effluent (E,).
1.65x106Pa 457 m/s pH8-10
240psig 15 ft/sec 52°C
(ACTUAL ELAPSED OPERATING TIME 470 HOURS)
fUNIT DOWN FOR CONTROLLER REPAIR.
-------
TABLE 7. ULTRAFILTRATION OF CAUSTIC BLEACH PLANT EFFLUENT, RUN U-1
Date,
1977
2-11
14
16
17
18
23
24
25
28
3-1
2
3
4
7
8
9
10
Elapsed
Operating
Time, hr.
3
68,
90*
115
139
1722
196
206
Unit down
230
254
278
302
326
398
422
446
470
Flux,
m3/m2/d
(gfd)
5.13
3.58
3.93
3.93
3.93
4.22
4.22
3.93
(126.0)
(87.8)
(96.6)
(96.6)
(96.6)
(103.5)
(103.5)
(96.6)
pH
Concen- Per-
trate meate
10.3
10.6
8.6
9.2
8.2
8.5
8.2
8.4
10.2
10.8
8.6
9.3
8.2
8.5
8.5
8.7
Conductivity
Concen- Per- % R
trate meate
5000
5500
4800
5000
5000 '
4100
4900
4300
3700
4000
2900
3000
2600
1500
2000
1400
26.0
27.3
39.6
40.0
48.0
36.6
59.2
67.4
Total
Concen-
trate
8555
24,804
37,505
38,340
48,941
49,730
45,981
57,152
Solids, p.p.m.
Per- % R
meate
3761
3513
2926
2764
2184
1605
1886
1304
56.0
85.8
92.2
92.8
95.5
96.8
95.9
97.7
over weekend
4.54
4.22
4.07
3.81
3.48
_
3.48
3.11
2.68
(111.5)
(103.5)
(100.0)
(93.5)
(85.3)
_
(85.3)
(76.3)
(65.9)
8.0
8.0
8.8
8.3
8.4
7.9
8.1
8.0
8.0
8.3
8.4
8.9
8.7
8.7
8.6
8.3
8.4
8.4
4600
5500
5800
6000
6200
6500
6200
6400
6400
2400
2500
3000
3200
3000
2500
2300
2800
2700
47.8
54.5
48.3
46.6
51.6
61.5
62.9
56.3
57.8
60,461
68,193
68,449
75,640
87,929
96,592
97,116
113,045
122,945
2214
2252
2789
2864
2683
2374
2360
3285
3189
96.0
96.7
95.9
96.2
96.9
97.4
97.6
97.1
97.4
o>
(continued)
Date,
1977
2-11
14
16
17
18
23
24»
25*
28
3-1
2
3
4
7
8
9
10
Elapsed
Operating
Time, hr.
3
68,
901
115
139,
1722
196
206
Unit down
230
254
278
302
326
398
422
446
470
Flux,
m3/m2/d
(gfd)
5.13 (126.0)
3.58 (87.8)
3.93 (96.6)
3.93 (96.6)
3.93 (96.6)
4.22 (103.5)
4.22 (103.5)
3.93 (96.6)
over weekend
4.54 (111.5)
4.22 (103.5)
4.07 (100.0)
3.81 (93.5)
3.48 (85.3)
3.48 (85.3)
3.11 (76.3)
2.68 (65.9)
Ash, p.p.m.
Concen- P«-
trate meate
3800
8504
5729
5931
8838
8857
9636
9861
10,874
12,411
14,799
15,437
16,435
18,086
17,311
18,796
18,313
2202
2220
1320
1460
1232
568
835
580
1130
1422
1822
1871
1602
1319
1120
1595
1585
%R
42.1
73.9
77.0
75.4
86.1
93.6
91.3
94.1
89.6
88.5
87.7
89.9
90.3
92.7
93.5
91.5
91.3
Volatile Solids,
Concen- Per-
trate meate
4755
16,300
31,776
32,409
40,103
40,873
36,345
47,291
49,587
55,782
53,650
60,203
71,494
78,496
79,805
94,249
104,632
1559
1293
1606
1304
952
1037
1051
724
1084
830
967
993
1081
1055
1240
1690
1604
p.p.m.
%R
67.2
92.1
94.9
96.0
97.6
97.5
97.1
98.5
97.8
98.5
98.2
98.4
98.5
98.7
98.4
98.2
98.7
^ Color, c. u
Concen- Per-
trate meate
28,500
162,500
250,000
287,500
345,000
325,000
385,000
390,000
500,000
554,000
560,000
476,000
614,000
645,000
680,000
850,000
890,000
326
1260
1800
1848
2030
1700
2050
1350
2200
3100
3450
4050
4120
3300
4400
4550
4600
%R
98.9
99.2
99.3
99.4
99.4
99.5
99.5
99.7
99.6
99.4
99.4
99.3
99.3
99.5
99.4
99.5
99.5
1 Unit down 2/15/77. Restarted.
2Unit down 2/19/77. Restarted.
-------
130
120
110,
100
2
"e
«n
E 90
X
at "
01 80
70
60
50
An
5.S
-
5.0
^4.5
~4.0
-I
3.5
-
3.0
"
2.5
""2.0
-
,
.
tit
0
* ^^ ^k ^^
.
-
1 I 1 1 1 1 1 1
15,000
30.000
45,000 60,000 75,000
TOTAL SOLIDS, ppm
90,000
105,000
120,000
Figure 33. Flux vs. total solids in concentrate, Run U-1.
-------
TABLES* FLUX VS. CONCENTRATE SOLIDS CONCENTRATION, RUN U-1
Date,
1977
2-11
14
16
17
18
23
24
25
28
3-1
2
3
4
7
8
9
10
Flux
m3/m2/d (gfd)
5.13
3.56
3.94
3.94
3.94
4.22
4.22
3.94
4.54
4.22
4.07
3.81
3.48
3.48
3.11
2.68
(126.0)
(87.8)
(96.6)
(96.6)
(96.6)
(103.5)
(103.5)
(96.6)
(111.5)
(103.5)
(100.0)
(93.5)
(85.3)
(85.3)
(76.3)
(65.9)
Total Solids, p. p.m.
8555
24,804
37,505
38,340
48,941
49,730
45,981
57,152
69,461
68,193
68,449
75,640
87,929
. 96,592
97,116
113,045
122,945
TABLE 9. PRESSURE DROP VS. CONCENTRATE SOLIDS
CONCENTRATION, RUN U-1
Date,
1977
2-11
14
15
16
17
18
23
24
25
28
3-1
2
3
4
7
8
9
id
Pressure Drop,
Pa (psig)
9.65 x 10s (140)
1.03 x106 (150)
1.07 x106 (155)
1.10x106 (160)
1.21 x106 (175)
1.21 x106 (175)
1.21 x 106 (175)
1.24 x 106 (180)
1.24 x106 (180)
1.24 x106 (180)
1.21 x 106 (175)
1.21 x 106 (175)
1.17x106 (170)
1.17 x106 (170)
1.24 x106 (180)
1.17 x106 (170)
1.17 x106 (170)
1.17 x106 (170)
Total Solids, p.p.m.
8553
24,804
_
37,505
38,340
48,941
49,730
45,981
57,152
60,461
68,193
68,449
75,640
87,929
96,592
97,116
113,045
122,945
66
-------
225
200
175
'55
^150
Q.
O
2125
DC
38 100
Q.
75
50
25
-
*
- . "
-^
-
-
1 I 1 1 1 I 1 f
360
330
300
270
240
210
180
150
120
90
60
30
15,000 30,000 45,000 60,000 75,000
TOTALSOLIDS.ppm
90,000
105,000
CD
Q-
120,000
Figure 34. Pressure drop vs. total solids concentration, Run U-1.
-------
membrane, when formed, gave lower rejections (13.5% conductivity) than desired at the end of the
formation step. This was possibly caused by the membrane having a thin spot at the point of con-
tact The high flux through spots could cause the operator to apply excess material and form a
thicker than necessary membrane (with lower flux) over the bulk of the tubes, while the high flux
and poor rejection indicates that membrane formation should continue.
It was decided to try this membrane with the waste water (Et) despite the fact that its quality
was less than desired. A 10 ym rating filter bag was used to filter the raw feed. The initial flux was
2.43 m3/m2/d (59.7 gfd) with a 31.6% conductivity rejection, but the permeate was not clear. The
flux dropped to 1.58 m3/m2/d (38.9 gfd) within 30 minutes and to 1.35 mVmVd (33.2 gfd) after
1 hour. In 3 hours the flux was 1.05 m3/m2/d (25.8 gfd), and after 3 days it was 0.82 m3/m2/d
(20.0 gfd) with a conductivity rejection of 27%. The low flux indicated a poor membrane, and
the unit was shut down temporarily. The membrane was inadvertently left dry for about 2 weeks,
and it was decided to try a Na2 C03 wash to see what improvement, if any, could be obtained. The
unit was restarted (Run U-2) with fresh waste water (Ej) after the wash, achieving an initial flux of
1.78 m3/m2/d (43.6 gfd) and a 42.9% conductivity rejection. After an hour of operation and a
power failure related shutdown, the flux had dropped to 1.06 m3/m2/d (25.9 gfd). As seen in
Figure 35 and Table 10, the flux decreased gradually from its initial value and leveled off at about
0.75 m3/m2/d (18.4 gfd). The rejections of color were good (94.3-99.1%), but the rejections of
solids were much poorer than those obtained in Run U-1. There were a number of power failure
related shutdowns during Run U-2, but these did not seem to have a significant impact upon the
unit's performance. The problems with membrane performance seemed to be related more to
problems with the formation step than with operational difficulties. On June 14, 1977, the unit
was shut down to perform a wash sequence in an attempt to improve the membrane performance. A
caustic wash (NaOH, pH = 11.8) and an acid wash (HN03, pH = 1.5) were performed, and an
attempt to add a second membrane produced a flux of only 2.20 m3/m2/d (54.1 gfd). Because
this flux was still lower than desired, a second wash sequence with extended washing times was
tried. As the flux did not improve, and new tube bundles were available, the unit was shut down,
and the bundles removed.
Unfortunately, the newer tube bundles proved to have even larger outside diameters than the
previous ones, and the bundles could not be loaded without breaking some of the tubes in the
outer ring. Thirteen bundles were finally loaded, but most of these bundles had several outer tubes
broken and stubbed off. It is certain that several bundles had tubes in contact with the piping, but
it was impossible to load them any other way. This problem with oversized tubes was corrected
later when the manufacturer replaced the worn extrusion die with which the tubes had been made.
A membrane was formed and, when started up (Run U-3, Figure 36 and Table 10) with the waste
water (Ei), it produced an initial flux of 3.98 m3/m2/d (97.6 gfd). However, in only 14 hours the
flux had dropped to 1.07 m3/m2/d (26.2 gfd) and in 42 hours to 0.70 m3/m2/d (17.1 gfd). This
very rapid flux decline, plus the poor rejections of the solids in the concentrate stream, indicated
another poor membrane, even though color rejections were good. The mill was shut down for a
Fourth of July holiday and mill maintenance work, and the unit was idle until July 11,1977. Fresh
waste water (E! ) was then added to the system, and on startup a flux of 1.30 m3/m2 /d (32.0 gfd)
was produced. The flux continued to drop for 2 days, at which time a wash sequence was perform-
ed. However, no improvement was obtained, and because a new module using porous carbon tubes
(Union Carbide Corp.) became available, this run was abandoned and the ceramic tube bundles were
removed. It was found that the old membranes could be removed from the ceramic tubes by heat-
ing to 105°C, but immersion in 1:1 HCI had no effect. It was observed that HCI-treated bundles did
not lose their membrane when heated, indicating that the acid may have altered the membrane
68
-------
100
75
IE 50
* 25
100
75
IE
#
25
100
75
IE go
* 26
100
75
a a
DO DO
TOTAL SOLIDS
ASH
B O
VOLATILE SOLIDS
60
45 -
30
15
60
45
30
15
12
COLOR
FLUX
2.0
1.0
CONDUCTIVITY
a
i I
j I
I I
PH
I
II I l
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500
TIME. HOURS
Figure 35. Run U-2. Ultrafiltration of caustic bleach effluent (Ej).
1.59x10° Pa 4.57 m/s
230psig 15 ft/sec
(ACTUAL ELAPSED OPERATING TIME 361 HOURS.)
52°C
PH9-11
-------
TABLE 10. ULTRAFILTRATION OF CAUSTIC BLEACH EFFLUENT, RUNS U-2 AND U-3
Date,
1977
Elapsed
Operating
Time, hr.
Flux,
m3/m*/d
(gfd)
pH
Concen- Per-
trate meats
Conductivity
Concen- Per- % R
trate maate
~ Total Solids, p.p
Concen- Per-
trate meate
.m.
%R
5/24
5/25
5/26
5/27
5/30
5/31
6/1
6/2
6/3
6/6
6/8
6/10
6/13
6/28
6/29
6/30
0.5,
181
40,
63 *
132
156
180
204
233.
249*
292.
336 l
361
0
14
42
1.78
1.11
0.97
0.50
0.75
0.76
0.77
0.76
0.67
0.74
0.79
0.73
0.64
3.98
1.07
0.70
(43.6)
(27.2)
(23.8)
(12.2)
(18.4)
(18.7)
(18.8)
(18.7)
(16.5)
(18.1)
(19.3)
(17.8)
(15.8)
(97.6)
(26.2)
(17.1)
11.0
10.2
10.1
9.9
9.6
9.7
9.4
9.6
9.4
9.1
9.6
9.2
9.2
Startup
10.6
10.5
11.2
10.1
10.1
9.8
9.5
9.6
9.3
9.5
9.3
9.1
9.4
9.0
9.0
10.8
10.7
2800
3800
3700
5000
5300
5400
5800
5900
6000
6000
5150
4500
5100
4900
5550
1600
2800
2500
3800
3800
3800
4300
4300
4500
4550
3400
2600
3400
3700
4250
42.9
26.3
32.4
24.0
28.3
29.6
25.9
27.1
25.0
24.2
34.0
42.0
33.3
24.0
23.0
3052
5163
5976
8095
13,550
13,568
15,421
17,114
18,585
18,542.
19,013
17,759
20,249
4801
7579
(Raw Waste)
1860
1744
2652
2734
2864
3628
3620
4255
4441
2931
2332
2942
2351
3390
64.0
70.8
67.2
79.8
78.9
76.5
78.5
77.1
76.0
85.6
86.9
85.5
37.0
55.3
vj
o
(continued)
Date,
1977
5/24
5/25
5/26
5/27
5/30
5/31
6/1
6/2
6/3
6/6
6/8
6/10
6/13
6/28
6/29
6/30
Elapsed
Operating
Time, hr.
0.5,
181
40.
63 1
132
156
»180
204
233.
249Z
292,
336*
361
0
14
42
Flux,
m3/m2/d
(gfd)
1.78 (43.6)
1.11 (27.2)
0.97 (23.8)
0.50 (12.2)
0.75 (18.4)
0.76 (18.7)
0.77 (18.8)
0.76 (18.7)
6.67 (16.5)
0.74 (18.1
0.79 (19.3)
0.73 (17.8)
0.64 (15.8)
3.98 (97.6)
1.07 (26.2)
0.70 (17.1)
Ash, p.p.m.
Concen- Per-
trata meate
%R
1821 (Raw Waste)
2671
2473
3274
5147
4635
5659"
6186
5674
6199
5960
4634
5464
Startup
1592
2486
1622
1364
1768
1888
1941
2093
1823
2882
2624
1393
1320
1500
1489
2007
39.3
44.8
46.0
63.3
58.1
63.0
70.5
49.2
57.6
76.6
71.5
72.5
6.5
19.3
Volatile Solids, p.
Concen- Per-
trate meate
1231
2492
3503
4821
8403
8933
9762
10,928
12,911
12,343
13,053
13,125
14,785
3209
5093
( Raw Waste)
238
380
884 ,
848
923
1535
1797
1373
1812
1538
1012
1442
862
1383
p.m.
%R
90.4
89.2
81.7
89.9
89.7
84.3
83.6
89.4
85.3
88.2
92.3
90.2
73.1
72.8
Concen
trate
3860
14,800
1 7,000
22,800
57.400
59,000
68,000
79,000
82,800
92,600
98,400
100,000
104,000
8000
16,300
"~ Color, CM.
Per-
meate
( Raw Waste)
171
148
426
592
780
980
4480
1216
1216
990
900
1190
463
690
%R
98.8
99.1
98.1
99.0
98.7
98.6
94.3
98.5
98.7
99.0
99.1
98.9
94.2
95.8
Temporary shutdown due to power failure. Restarted.
'Shut down during weekend due to power failure. Restarted.
2,
-------
IIAJ
75
^ 50
25
100
75
"= 50
ae
25
100
75
* 50
25
100
75
QC
25
100
"5, 75
3s0
"" 25
100
75
oc
25
12
9
z 6
a
3
TOTAL SOLIDS
- ASH
"
_ 9
TOTAL ORGANICS (BY COMBUSTION)
* *
COLOR
FLUX
CONDUCTIVITY
PH
I 1 I I I I I I I
4.0
3.0
2.0
1.0
0 5 10 15 20 25 30 35 40 45 50
TIME, hours
Figure 36. Run U-3. Ultrafiltration of caustic bleach effluent (Ej).
Continuation of Run U-2 with new membrane and tubular supports.
1.59x106Pa 4.57 m/s 52°C
230psig 15 ft/sec
71
-------
material in some way. Because these regenerated tubes were oversized, they were not reinstalled
and the following Run U-4 was performed using only the carbon tube module.
The carbon tube module was built by Oak Ridge National Laboratory and consisted of a stainless
steel shell containing Luciteblocks with seven holes drilled through them. The holes were aligned to
allow the carbon tubes to pass through and were of the proper diameter to allow free flow over the
tube exterior. The tubes were suspended away from the Lucite blocks with two spacer rings
which were also cut to allow free flow past them. One end of each tube was sealed, and the other
was cemented into a stainless steel threaded bushing which screwed into the end plate of the shell.
This allowed passage of the permeate from the interior of the tube out of the shell. Each tube was
1.02 cm (0.4 in.) outside diameter and a nominal 127 cm. (50 in.) long, and with seven tubes in-
stalled in the module the total membrane surface area was 0.285 M2 (3.07 sq. ft.). The module was
mounted in place of the UCARSEP module and incorporated into the system during membrane
formation and operation in the same manner as ceramic tubes. This unit's advantage was that it
could be run at high pressures, up to 6.894x106 Pa (1000 psig) and higher, allowing its possible use
as a reverse osmosis membrane support as well as an ultrafiltration support. The membrane forma-
tion and operation parameters were the same as those of the ceramic tube bundles, except that the
concentrate flow rate was raised slightly to achieve the surface velocity of 4.57 m/s (15 ft/sec).
The first membrane was applied to the carbon tubes on July 22, 1977, and the unit started up
using fresh waste water (Ex). The raw feed was filtered through a 5 vim rating filter bag and a slimi-
cide, Metasol D3TA (Merck Chemical Co.), was added to prevent biological growths. The initial
flux was 1.61 m3/m2/d (39.6 gfd) with 19.7% conductivity rejection. Over the weekend, the flux
dropped to 0.92 m3/m2/d (22.5 gfd) and stayed at this level for 12 days. Unfortunately, there are
no chemical data for this period because the samples were inadvertently discarded. The unit was
shut down when the raw feed line from the bleach plant became clogged, and after a water rinse,
the initial flux on startup with fresh Ej was only 0.98 m3/m2/d (24 gfd). The unit operated irre-
gularly for several days because of controller problems and was shut down until late September. On
September 26,1977, awash sequence was initiated in preparation fora new membrane formation. Dur-
ing the acid wash, total power was lost, and the entire unit was left filled with 0.5 JW HN03 until
October 10,1977. When restarted with HN03 solution, the flux was initially 2.47 m3/m2/d (60.7
gfd), rose to 3.46 m3/m2/d (85 gfd) in about 1 hour, and after 2 more hours had dropped to 2.90
m3/m2/d (71.2 gfd). The unit was drained, rinsed, and an additional layer of hydrous Si(IV) oxide
applied. Upon restarting with fresh waste water on October 12,1977, the initial flux was 1.41 /
m3/m2/d (34.7 gfd). No Metasol was used in this or subsequent runs because it was theorized that
the slimicide was adversely affecting the flux. The data from this run (U-4) are shown in Figure 37
and Table 11. The flux dropped gradually until October 17,1977 when the'leaking tank caused
the level controller to add excess raw waste. The dilution and slightly lower solids concentration prob-
ably contributed to the flux increase. However, the flux and the total solids concentration rose
even higher on October 18, 1977. The unit was found shut down on October 19,1977 and re-
started, but on October 20, 1977 the tank was found overflowing because of a jammed feed valve.
The concentrate in the system was drained out and the unit refilled and restarted. The initial flux
was 1.91 m3/m2/d (46.8 gfd), but after 7 hours the flux had risen to 2.02 m3/m2/d (49.6 gfd). The
next day the flux held steady and rose slightly over the weekend.
The flux rates obtained were higher than those in Runs U-2 and U-3 but less than those of Run
U-1. The soaking in HNO3 and additional hydrous Si(IV) oxide seemed to improve the flux some-
what, although little change was noted in the rejection of the constituents. Overall color rejection
72
-------
IUU
75
C 50'
as
25
100,
75
* "
25
100
75,
K 50<
25<
100
7&
CC so
25
40
30'
* 2°
10
10.0
'
7.5
x 5.0
Q.
2.5
100
B 75
O)
$ ",
s! 25
0 o oo o o o
O SODIUM
o Q DO a O TOC
D
DO ° a DO a
0 D VOLATILE SO LIDS
_ O COLOR
w O
a D a o D o COD °
f 0 TOTAL SO LIDS
00 ° ° o ° CHLORIDES 0
D a D D
a
D TOTAL SULFUR
go 00 00OASH Q
.
* CONDUCTIVITY
-
L * * *
pH
. * FLUX *
4.0
3.0,
2.0 E
1.0
20 40 60 80 100
120 140 160
TIME, hours
180 200 220 240 260 280
Figure 37. Run U-4. Ultrafiltration of caustic bleach effluent (En), carbon tube only.
1.52x106Pa
220 psig
4.57 m/s
15 ft/sec
52°C
pH 8-10
73
-------
TABLE 11. ULJRAFILTRATION OF CAUSTIC BLEACH EFFLUENT USING CARBON TUBE MODULE, RUN U-4.
Date,
1977
10/12
10/13
10/14
10/17
10/18
10/20-
10/21
10/24
Elapsed
Operating
Time, hr.
8-1/3
21,
401
123,
1353
209
287
Flux,
m3/m2/d
(gfd)
1.45 (35.7)
1.33 (32.7)
1.30 (32.0)
1.56 (38.3)
1.89 (46.5)
2.02 (49:6)
1.99 (48.9)
2.21 (54.3)
pH Conductivity Total Solids,
Concen- Per- Concen- Per- % R Concen- Per-
trate meate trate meate trate meate
8.75 8.85
8.50 8.40
8.60 8.55
9.25 9.15
8.86 8.70
8.35 8.45
8.60 8.50
7.90 7.50
4500 -3100
4800 3600
4900 3700
4800 3350
5300 4250
4050 3000
4350 3350
3800 3050
31.1
25.0
24.5
30.2
19.8
25.9
23.0
19.7
4647 2187
5689 2504
6804 2931
6157 2548
8052 3456
4482 2129
5690 2308
4787 2207
p.p.m.
%R
52.9
56.0
56.9
58.6
57.1
52.5
55.9
53.9
Ash, p.p.
Concen- Per-
trate meate
1957 1147
2806 1820
3502 2190
3406 2144
3978 2540
2552 1744
2939 1918
2574 1784
m. Chloride, p.p.m.
% R Concen- Per- % R
trate meate
41.4 1197
35.1 1175
37.5 1259
37.1 1221
36.2 1361
31.2 1039
34:7 1072
30.7 896
802
993
1092
978
1193
844
896
806
33.0
15.5
13.3
19.9
12.3
23.1
16.4
10.0
(continued)
Date,
1977
10/12
10/13
10/14
10/17
10/18
10/20
10/21
10/24
Elapsed
Operating
Time, hr.
8-1/3
21 1
40*
123^
135|
209
287
m An /d
(gfd)
1.45 (35.7)
1.33 (32.7)
1.30 (32.0)
1.56 (38.3)
1.89 (46.5)
2.02 (49.6)
1.99 (48.9)
2.21 (54.3)
' ' Color, c.u. *-' Sodium, p.p.m
Concen- Per- % R Concen- Per-
trate meate trate meate
7450 1 13 98
10,440 6940 41
12,450 7650 39
13,850 5060 63
10,200 150 98
9500 179 98
10,200 160 98
9500 1 79 98
.5 1001.2
.1 1295.3
.4 1409.6
.5 1375.6
.5 1726.5
.1 982.7
.5 1168.4
.1 977.0
383; 2
763.2
854.1
778.0
994.6
677.9
664.6
600.0
., as Na
%R
61.7
41.1
39.4
43.4
42.4
31.0
43.1
38.6
5-Day BOD, p.p.m.
Concen- Per- %
trate meate
370
COD, p.p.m.
R Conden- Per- % R
trate meate
88 76.2 1522
2499
- - 4060
- - 3407
- - 5130
- - 2251
- - 3291
- - 2710
440
581
510
387
653
286
407
369
71.1
76.8
87.4
88.6
87.3
87.3
87.6
86.4
(continued)
Date,
1977
10/12
10/13
10/14
10/17
10/18
10/20
10/21
10/24
Elapsed
Operating
* Time, hr.
8-1/3
21 1
40*
123,
135|
209
287
m3/m2/d
(gfd)
1.45 (35.7)
1.33 (32.7)
1.30 (32.0)
1.56 (38.3)
1.89 (46.5)
2.02 (49.6)
1.99 (48.9)
2.21 (54.3)
TOC, p.p.m.
Concen- Per-
trate meate
775
1080
1463
1170
1890
904
1186
1088
57
94
150
150
277.4
128.2
184.0
170.2
%R
92.6
91.3
89.7
87.2
85.3
85.8
84.5
84.4
Total Sulfur,
Concen- Per-
trate meate
12
16
26
17
22
14
17
3.0
6.0
6.0
3.5
7.0
3.0
4.0
p.p.m.
%R
75.0
37.5
76.9
79.4
68.2
78.6
76.5
Volatile Solids,
Concen- Per-
trate meate
2690
2883
3302
2751
4074
1930
2751
2213
1040
684
741
404
916
385
590
423
p.p.m.
%R
61.3
76.3
77.6
85.3
77.5
80.1
78.6
80.9
1
Concentrate tank leaking. Some loss of concentrate feed.
2 Unit shut down on 10/19/77 due to clogged feed line. Restarted.
Level control failure. Drained tank and restarted with fresh feed.
-------
in Run U-4 was good, but it was not as high as with the ceramic tubes and the "good" membrane
used in Run U-1. The fair to good rejections of volatile solids, TOC, total sulfur, and COD, and the
low rejections of sodium ash and chlorides indicate that this ultrafiltration process for treating
bleach plant wastes is a valid concept, even when operation is not optimal. The large organic mole-
cules are being retained, while the chlorides are being passed.
The unit was found shutdown on October 25, 1977 because of a power failure. As ceramic
tube bundles were now available, they were installed after a water rinse. The unit was filled with a
NaOH solution (pH = 12.9) and allowed to run overnight. On stopping the wash and rinsing the
unit out with mill supply water, it was observed that the flux from the ceramic tubes had dropped
dramatically, almost as if a membrane had been formed. Apparently the materials that washed out
of the piping were deposited on the porous tubes, in a similar manner to that of studies on lignin
membranes. An acid wash lasting over 3 hours had little effect because the flux during the deioniz-
ed water rinse that followed was only 4.36 m3/m2/d (107 gfd), as compared with the high fluxes
(81.48 m3/m2/d [2000 gfd]) expected for new tube bundles. Because no other tube bundles were
then available, a membrane was applied on the ceramic tubes. The carbon tube bundle was isolated
during the membrane formation step. The flux for the ceramic tubes was 4.25 m3/m2 /d (104.3 gfd)
at the end of the hydrous Si(IV) oxide layer formation. The unit was filled with fresh waste water
(E! ) and started up, using both the carbon tubes and ceramic tube bundles. The data for the
ceramic tubes is given as Run U-5 and for the carbon tubes as Run U-6; however, the runs occurred
simultaneously and were treating the same feed.
The data for Run U-5 are given in Figure 38 and Table 12. The initial flux on startup was 2.75
m3/ma/d (67.4 gfd), but it dropped slightly within 30 minutes to 2.59 m3/m2/d (63.6 gfd) and
leveled out the next several days at slightly below 2.04 m3/m2/d (50 gfd). The unit was shutdown
on November 4,1977 because of poor permeate quality. The fluxes from the individual bundles
ranged from 1.36 to 2.72 m3/m2/d (33.4 to 66.7 gfd). The color removal from different bundles
varied widely also, but no correlation was evident between flux and color removal. The bundles
with the lowest fluxes did not give higher color rejections than those with the highest fluxes. The
unit was washed using the regular procedure and left for the weekend with the 0.5 N HNO3 (pH =
1.1) solution in it. A layer of hydrous Si(l V) oxide was applied to both the ceramic and carbon
tubes, and the unit restarted with fresh waste water (E! ). The initial flux was 2.66 m3/m2/d (65.4
gfd), but the flux rose in 2 hours to 3.02 m3/m2/d (74.2 gfd). The unit ran on an irregular schedule
for several days because of power failures, tank leaks, and loss of feed. On November 16,1977, the
unit was producing very dark permeates, and the unit was shut down until November 28,1977 for
cleanup and new membrane.
The additional membrane produced an initial flux of 2.77 m3/m2/d (68.1 gfd), but within 1
day the flux rose to 3.47 m3/m2/d (85.1 gfd). The permeates were dark brown in appearance, but
the data in Table 12 show that the unit was achieving good rejections of color and organics while
progressively concentrating the total solids. This dark color in the permeate and the shortage of
time forced the shutdown of the unit on December 7,1977. The data shown in Table 12 indicate
that this last membrane was a fairly good one because high fluxes 4.07 m3/m2/d (100 gfd) were
possible with good rejections. While the overall performance of the membrane was not quite as
good as Run U-1, the unit's performance could be improved under less than ideal conditions. This
ability to adjust to problems encountered and to regenerate the membranes is perhaps the greatest
advantage to using dynamically formed membranes. The last membrane addition resulted in im-
proved rejections and higher fluxes.
75
-------
1UU
75
C 50
* 25
75
1C SO
* 25
inn
75
C 50
* 25
inn
75
C 50
* 25
45
? 30
* 15
12
9
IfiO
SIM
SS
11.
000 0 0 0° °0
°D "on
D
n
°B8 °H 80 8"
0 o o Q o o
o o jj a S on "
^ o o » "o oo 0°
.
*
r B o 0 o a ° fl
° o ° o
_
... . .
.
- . . . .
-
-
* *
1
i
1
1
s^;
ta
H
§$$^
ft
i
^
1
> ° ° ^ " Q
a " °
° a SODIUM
1 » TOC
' B "
0
a VOLATILE SOLIDS
o COLOR
' °0 0°
a COD
o TOTAL SOLIDS
I i 8
E» fl H . BODS
o ASH
o TOTAL SULFUR
* CONDUCTIVITY
.
. pH
' FLUX _
6.0
4.0;
2.0
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 | 660 680 700 720 740 760 780 800 820 840 860
TIME, hours »New membrane added
Figure 38. Run U-5. Ultrafiltration of caustic bleach effluent (E,) using ceramic bundles.
1.59x10" Pa 4.57 m/s 52°C
230psig 15 ft/sec pH8-10
-------
TABLE 12. ULTRAFILTRATION OF CAUSTIC BLEACH EFFLUENT USING CERAMIC MEMBRANE SUPPORT, RUN U-5
Date,
1977
11/1
11/2
11/3
11/4
11/7
11/8
11/10
11/11
11/14
11/15
11/28
11/29
11/30
12/2
12/5
12/7
Elapsed
Operating
Time, hr.
1
23
51,
711
145
170,
220-4
244.
309d
3394
652
676r
7015
749
820
867
.flux,
m3/m2/d
(gfd)
2.59 (63.6)
1.93 (47.3)
1.89 (46.4)
2.01 (49.4)
2.92 (71.8)
2.66 (65.4)
3.12 (76.7)
3.02 (74.2)
5.04 (123.6)
4.80 (117.9)
2.77 (68.1)
3.47 (85.1)
3.-74 (91.7)
3r74 (9V.7)
4.04 (99.3)
4.04 (99.3)
pH
Concert- Per-
trate meate
10.10 10.10
9.50 0.50
9.60 9.60
8.85 8.80
10.50 10.60
9.65 9.60
9.75 9.80
8.80 8.80
9.35 9.20
8.80 8.60
9.60 9.45
9.55 9.60
9.65 9.40
9.10 9.10
8.90 9.10
9.85 10.10
Conductivity
Concen- Per % R
trate meate
4800
5250
5850
5200
3500
4100
6100
6700
4850
3500
4650
4100
4350
5100
4500
5200
3800
4250
4600
4150
2400
2600
4400
5500
3750
2700
3500
4100
3250
3600
2500
3200
20.8
19.0
21.4
20.2
31.4
36.6
27.9
17.9
22.7
22.9
24.7
_
25.3
29.4
44.4
38.5
Total Solids, p.p.m.
Concen- Per- % R
trate meate
5084
12,150
21,434
21,736
3349
10,127
20,739
32,576
14,869
5944
5484
17,711
21,498
36,636
41,277
2388
3597
4413
3910
1332
1809
4229
5630
2995
2104
2694
3653
4701
3169
2375
52,036 3039
53.0
70.4
79.4
82.0
60.2
82.1
79.6
82.7
79.9
64.6
50.9
79.4
78.1
91. -4
94.3
94.2
Ash, p.p.m.
Concen- Per- % R
trate meate
2910
4311
5745
5211
1668
2933
5525
8404
4043
2328
2946
5557
5588
7717
7922
10,791
1808
2363
2726
2161
931
1149
2848
3662
1875
1374
2006
2452
1598
1859
1316
1744
37.9
45.2
52.6
58.5
44.2
60.8
48.5
56.4
53.6
41.0
31.9
55.9
71.4
75.9
83.5
83.8
Chloride, p.p.m.
Concen- Per- % R
trate meate
975
941
753
772
1050
822
571
745
658
934
15.7
39.3
1.1
14.8
11.0
(continued)
Date,
1977
11/1
11/2
11/3
11/4
11/7
11/8
11/10
11/11
11/14
11/15
11/28
11/29
11/30
12/2
12/5
12/7
Elapsed
Operating
Time, hr.
1
23
51.
71 *
145
170,
220
244,
309 J
3394
652
676,-
701 5
749
820
867
Flux,
m3/m2/d Concen-
(gfd) trate
2.59 (63.6) 7450
1.93 (47.3) 41,750
1.89 (46.4) 84,000
2.01 (49.9) 48,000
2.92 (71.8) 2600
2.66 (65r4) 39,100
3.12 (76.7) 89,000
3.02 (74.2) 159,000
5.04 (123.6) 61,000
4.80 (117.9) 14,600
2.77- (68.1) 10,900
3.47- (85.1) 83,000
3.74- (91.7) 110,500
3.74- (91. 7) 207,000
4.04 (99.3) 232,000
4.04 (99.3) 291,000
Color, c.u.
Per-
meate
243
890
1800
1800
68
510
2040
4100
1420
482
273
1610
1730
2840
2750
4500
%R
96.7
97.9
97.9
96.3
97.4
98.7
97.7
97.4
97.7
96.7
97.5
98.1
98.4
98.6
98.8
98.5
-Sodium, p.p.m.,
Concen- Per
trate meate
1051.6
1826
2429
1924
418
1530
2944
3973
2136
958
1129
2411
2833
4019
4749
666
860
773
538
402
508
1157
1505
802
537
729
881
652
834
731
asNa
%R
36.7
52.9
68.2
72.0
3.8
66.8
60.7
62:1
62.5
44.0
35.4
63".5
77.0
79:3
_
84.6
5-Day BOD, p.
Concen- Per-
trate meate
1320
_
2100
_
-_
_
4400
1700
810
560
2600
1800
5000
4700
3600
_
300
_
350
_
_
-i.
590
240
220
220
340
300
360
380
360
p.m.
%R
_
77.3
_
83.3
_
_
_
86.6
85.9
72.8
60.7
86.9
83.3
92.8
91.9
90.0
COD, p.p.m
Concen- Per-
trate meate
2614
9934
19,600
29,375
1700
8400
22,500
39,000
15,600
4425
3100
15,500
26,875
49,400
52,800
66,800
465
980
1360
1400
220
490
1310
2050
730
540
460
1135
1380
1375
1330
1550
%R
82.2
90.1
93.1
94.2
87.1
94.2
94.2
94.7
95.3
87.8
85.2
92.7
94.9
97.2
97.5
97.7
Unit shut down because of poor permeate quality. Additional SKI V) oxide membrane added, restarted.
2Unlt down all day (11/9/77) due to power failure. Restarted 11/10/77.
Tank leak forced shutdown. Restarted 11/15/77.
A
Unit shut down to apply new membrane. Startup delayed until 11/28/77 due to tank and other repairs.
Tank found overflowing. Raw feed valve jammed open and dilution of concentrate occurred, restarted.
(continued)
-------
TABLE 12 (continued)
vj
00
Date,
1977
11/1
11/2
11/3
11/4
11/7
11/8
11/10
11/11
11/14
11/15
11/28
11/29
11/30
12/2
12/5
12/7
Elapsed
Operating
Time, hr.
1
23
52 1
71.
145
170,
220^
244,
309 5
339
652
6764
701 4
749
820
867
Flux,
m3/m2/d
(gfd)
2.59 (63.6)
1.93 (47.3)
1.89 (46.4)
2.01 (49.4)
2.92 (71.8)
2.66 (65.4)
3.12 (76.7)
3.02 (74.2)
5.04 (123.6)
4.80 (117.9)
2.77 (68.1)
3.47 (85.1)
3.74 (91.7)
3.74 (91.7)
4.04 (99.3)
4.04 (99.3)
TOC, p.p.m. *
Concert- Per- % R
trate maate
995
3505
8420
8750
690
3575
5550
13,663
5472
1650
1160
6921
10,160
16,814
19,137
94.075
181.2
306
555
540
66
188
555
953
425
220
165
513
502
595
605
765
81.8
91.3
93.4
93.8
90.4
94.7
90.0
93.0
92.1
86.7
85.8
92.6
95.1
96.5
97.4
96.8
Total Sulfur, p.p.m.
Concen- Per- % R
trate meate
17
46
76
33.1
8.9
27.3
42.5
56.9
27.3
13.2
23.1
44.0
40.1
45.7
14.0
5.0
13.0
20.0
11.0
6.6
8.5
11.2
16.0
7.9
4.1
8.6
14.3
12.4
12.9
1.0
70.6
71.7
73.7
66.8
25.8
68.9
73.6
71.9
71.1
68.9
62.8
67.5
69.1
71.8
92.9
Volatile Solids, p.p.m.
Concen- Per- % R
trate meate
2174
7839
1 5,689
16,525
1681
7194
15,214
24,172
10,826
3616
2538
12,154
15,910
28,919
33,355
41 ,240
580
1234
1687
1330
401
660
1381
1968
1120
730
638
1201
3103
1310
1059
1295
73.3
84.3
89.3
89.8
76.2
90.8
90.9
91.9
89.7
79.8
74.9
90.1
80.5
95.5
96.8
96.9
Unit shut down because of poor permeate quality. Additional Sl(l V) oxide membrane added, restarted.
Unit down all day (11/9/77) due to power failure. Restarted 11/10/77.
Tank leak forced shutdown. Restarted 11/15/77.
4
Unit shut down to apply new membrane. Startup delayed until 11/28/77 due to tank and other repairs.
Tank found overflowing. Raw feed valve jammed open and dilution of concentrate occurred, restarted.
-------
The series of events is identical for Run U-6 (the carbon tube module) as for Run U-5. The
initial flux on startup was 1.87 m3/m2/d (46.0 gfd), but this dropped to 1.54 m3/m2/d (37.9 gfd)
in 30 minutes. The flux rose slightly and remained stable for several days. The permeate color
became dark, and the unit was shut down in preparation for an additional membrane layer. The
unit was left full of 0.5 N HNO3 (pH = 1.1) over the weekend, and a layer of hydrous Si(IV) oxide
was applied. The initial flux upon restarting with fresh waste water (Ex) was 1.45 m3/m2/d (35.7
gfd), with an increase to 1.84 m3/m2/d (45.3 gfd) in 2 hours. The unit ran on an irregular schedule
for several days because of tank repairs and power failure, but the flux gradually increased. On
November 16,1977, the permeate quality appeared to be poor, so the unit was shut down for
cleanup and a new membrane formation. The startup with the new membrane was delayed until
November 28,1977 by necessary repairs.
The initial flux on startup was 1.80 m3/m2/d (44.3 gfd), and the flux gradually rose over the
rest of the run. The rejections of organics and solids improved slightly with the addition of the last
membrane, as seen in Figure 39 and Table 13. The carbon and ceramic tube supports did not
appear to have any distinct advantage over one another with regard to rejections, but the ceramic
supports gave slightly higher fluxes. As with the ceramic tubes, flux and rejection performance was
increased by adjustments to the dynamic membranes.
79
-------
75
5 50
* 25
inn
75,
C 50
* 25
irm
75
CC 50
* 25
100
75
* 50
* 25
45
15
00 13
0 1Z
9
X 6
3
1 120
X 80
3 40
U.
0 0 0 o ° 0° 00
° 0
.. O " O DO O
O
;
0 ° 0° ° 5 ° 0
0° 0 00 o
. °
; D o ° 0 o ;
0 ° 0 ° 0 ° 0
*
. . .
-
1 1 1 I I 1 1 ' 1 1 1 1 1 1 1 1 1 1
m
t
^
n
$$$s
"^sSfe
1
1
jss§;
5^Ss
I
^^
^^
1
I5w
^
x^
p
00 » " °
:
a a SODIUM
o TOC
8
a
o VOLATILE SOLIDS
o COLOR
i a 8 o *
' 0 "
o COD
o TOTAL SOLIDS
8 .
8 g B & BODS
o D TOTAL SULFUR
o ASH
.
* CONDUCTIVITY
, , »
f>H
.
FLUX
6.0
4.0 -y
2.0 C
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 * 660 680 700 720 740 760 780 800 820 840 860
TIME, hours ,
, Figure 39. Run U-6. Ultrafiltration of caustic bleach effluent (Ej) using carbon tube module.
1.59x106Pa
230 psi
4.57 m/s
15 ft/sec
52°C
PH8-10
*New membrane applied.
-------
TABLE 13. ULTRAFILTRATION OF CAUSTIC BLEACH EFFLUENT USING CARBON MEMBRANE SUPPORT, RUN U-6.
Date,
1977
11/1
11/2
11/3
11/4
11/7
11/8
11/10
11/11
11/14
11/15
11/28
11/29
11/30
12/2
12/5
12/7
Elapsed
Operating
Time, hr.
1
23
51,
71 l
145
170,
220Z
244,
309 J
339d
562
676c
701 5
749
820
867
Flux,
m3/mJ/d
(gfd)
1.54 (37.9)
1.68 (41.3)
1.64 (40.4)
1.73 (42.5)
1.83 (44.8)
2.16 (53.1)
2.53 (62.0)
2.53 (62.0)
3.44 (84.5)
2,53 (62.0)
1.80 (44.3)
2.91 (71.5)
3.15 (77.4)
3.15 (77.4)
3.44 (84.5)
3.61 (88.5)
PH
Concen- Per-
trate meate
10.10
9.50
9.60
8.85
10.50
9.65
9.75
8.80
9.35
8.80
9.60
9.55
9.40
9.10
8.90
9.85
10.10
9.50
9.65
8.80
10.60
9.65
9.85
8.80
9.30
8.70
9.70
9.65
9.40
9.15
9.05
10.05
Conductivity
Concen- Per- % R
trate meate
4800
5250
5850
5200
3500
4100
6100
6700
4850
3500
4650
5500
4800
5100
4500
5200
4050
4550
5250
4150
2600
2750
4400
5400
3700
2700
3750
4350
3500
3700
2900
3600
15.6
13.3
10.3
20.2
25.7
32.9
27.9
19.4
23.7
22.9
19.4
20.9
27.1
27.5
35.6
30.8
Total Solids, p.p.m.
Concen- Per- % R
trate meate
5084
12,150
21,434
21,736
3349
10,127
20,739
32,576
14,869
5944
5484
17,711
21,498
36,636
41,277
52,036
2765
4053
5096
3910
1477
1981
4159
5817
3189
2082
2724
4162
3080
3691
2656
3754
45.6
66.6
76.2
82.0
55.9
80.4
80.0
82.1
78.6
65.0
50.3
76.5
85.7
89.9
93.6
92.8
Ash, p.p.m.
Concen- Per- % R
trate meate
2910
4311
5745
5211
1668
2933
5525
8404
4043
2328
2946
5557
5588
7717
7922
10,791
2099
2596
3035
2381
1072
1296
2871
3857
2068
1418
2085
2780
1972
2287
1543
2378
27.9
39.8
47.2
54.3
35.7
55.8
48.0
54.1
48.9
39.1
29.2
50.0
64.7
70.4
80.5
78.0
Chloride, p.p.m.
Concen- Per- % R
trate meate
975
941
_
772
1050
"
932
549
720
975
~
4.4
41.7
6.7
7.1
~
(continued)
oo
Date,
1977
11/1
11/2
11/3
11/4
11/7
11/8
11/10
11/11
11/14
11/15
11/28
11/29
11/30
12/2
12/5
12/7
Elapsed
Operating
Time, hr.
1
23
51,
711
145
170,
2202
244..
309
3394
652
676c
701s
749
820
867
Flux,
m3/m2/d
(gfd)
1.54. (37.9)
1.68 (41.3)
1.64 (40.4)
1.73 (42.5)
1.83 (44.8)
2.16 (53.1)
2.53 (62.0)
2.53 (62.0)
3.44 (84.5)
2.53 (62.0)
1.80 (44.3)
2.91 (71.5)
3.15 (77.4)
3.15 (77.4)
3.44 (84.5)
3.61 (88.5)
^_ ___
Concen-
trate
7450
41,750
84,000
48,000
2600
39,100
89,500
1 59,000
61,000
14,600
10,900
83.000
110,500
207,000
232,000
291,000
Color, c.u.
Per-
meate
154
900
2060
1800
53
301
1164
2540
820
103
102
1540
1540
2840
2410
2750
Sodium, p.p.m..
%R
97.9
97.8
97.6
96.3
98.0
99.2
98.7
98.4
98.7
99.3
99.1
98.1
98.6
98.6
99.0
99.1
Concen-
trate
1051
1826
2429
1924
1530
2944
3973
2136
958
1129
2411
2833
4019
_
4749
Per-
meate
716
976
879
665
_
571
1268
1649
853
579
365
1035
794
891
_
853
asNa
% R
31.9
46.6
63.8
65.4
_
62.7
56.9
58.5
60.1
39.6
67.7
57.1
72.0
77.8
_
82.0
5-Day BOD, p.p.m.
Concen-
trate
1320
2000
2100
_
_
2100
4400
1700
810
560
2600
1800
5000
4700
3600
Per-
meate
270
390
420
_
_
310
670
310
160-
145
420
380
400
350
390
%R
79.5
80.5
80.0
_
85.2
84.8
81.8
80T.3
74.1
83.9
78.9
92.0
92.6
89.2
COD, p.p.m.
Concen-
trate
2614
9934
19,600
29,375
1700
8400
22,500
39,000
15,600
4425
3100
15,500
26,875
49,400
52,800
66,800
Per-
meate
445
1060
8075
1575
180
440
1185
1860
895
37.0
365
1265
1380
1380
1460
1650
%R
83.0
89.3
58.8
94.6
89.4
94.8
94.7
95.2
94.3
91.6
88.2
91.8
94.9
97.2
972
97.5
Unit shut down because of poor permeate quality. Additional SKI V) oxide membrane added, restarted.
Unit down all day (11/9/77) due to power failure. Restarted 11/10/77.
Tank leak forced shutdown. Restarted 11/15/77.
A
Unit shut down to apply new membrane. Startup delayed until 11/28/77 due to tank and other repairs.
Tank found overflowing. Raw feed valve jammed open and dilution of concentrate occurred, restarted.
(continued)
-------
TABLE 13 (continued)
Date,
1977
11/1
11/2
11/3
11/4
11/7
11/8
11/10
11/11
1 1/14
11/15
11/28
11/29
11/30
12/2
12/5
12/7
Elapsed
Operating
Time, hr.
1
23
61,
71 1
145
170,
220-*
244,
3093
3394
652H
676.
7015
749
820
867
Flux,
m3/m2/d
(gfd)
1.54 (37.9)
1.68 (41.3)
1.64 (40.4)
1.73 (42.5)
1.83. (44:ar
2.16 (53.1)
2.53 (62.0)
2.53 (62.0)
3.44 (84.5)
2.53 (62:0)
1.80 (44.3)
2.91 (71.6)
3.15 (77:4)
3.14 (77.4)
3.44 (84.5)
3.61 (88.5)
TOC, p.p.m.
Concen- Per- % R
trate meate
995
3505
8420
8750
690
3675
5550
13,663
6472
1650
1160
6921
10,160
6814
19,137
24,075
171.6
317.4
684.0
640.0
66.0
148.0
556.0
609.0
371.0
136.0
158.0
585.0
578.0
706.0
559
700
82.8
90.9
91.9
92.7
90.4
95.9
90.0
95.5
93.2
91.8
86:4
91.5
94.3
95.8
97.1
97.1
Total Sulfur, p.p.m.
Concen- Per- % R
trate meate
17
46
76
33.1
8.9
27.3
42.5
56.9
27.3
13.2
23.1
44.0
40:1
45.7
14.0
*~*
5
18
26
10.9
4.6
7.8
13.7
17.0
9.7
4.0
8.0
19.0
16.0
18.1
1.0
58.8
60.9
65.8
67.1
48.3
71.4
67.8
70.1
69.5
69.7
65.4
56.8
61.1
60.4
92.9
Volatile Solids, p.p.m.
Concen- Per- % R
trate meate
2174
7839
15,689
16,525
1681
7194
15,214
24,1 72
10,826
3616-
2538
12,154
15,910
28,919
33,355
41,240
666
1457
2161
1529
405
670
1288
1960
1121
664
639
1382
1108
1404
1113
1356
69.4
81.4
86.2
90.8
75.9
90.7
91.5
91.9
89.7
81.6
74.8
88.6
93.0
95.2
96.7
96.7
Unit shut down because of poor permeate quality. Additional Sid V) oxide membrane added, restarted.
Unit down all day (11/9/77) due to power failure. Restarted 11/10/77.
Tank leak forced shuidown. Restarted 11/15/77.
4
Unit shut down to apply new membrane. Startup delayed until 11/28/77 due to tank and other repairs.
Tank found overflowing. Raw feed valve jammed open and dilution of concentrate occurred, restarted.
-------
SECTION 7
DISCUSSION OF RESULTS
REVERSE OSMOSIS UN IT
The pilot plant unit did not prove reliable enough to provide long-term operation data, but the
trials did produce useful information in spite of the mechanical difficulties. The major parameters
of interest are the fluxes obtained and the rejections of various constituents achieved by this type
of membrane system. Most commercially available reverse osmosis units are capable of removing
organic and inorganic solutes, but at low flux rates, usually 0.407 to 0.611 m3/m2/d (10-15 gfd).
Most types have limitations on temperature and pH as well. The dynamic membrane system is
potentially useful for waste-water treatment because higher fluxes and at least equivalent removals
are possible.
Run H-1 started with an initial flux of 2.35 m3/m2/d (57.6 gfd), and the flux declined to 1.31
m3/m2/d (32.2 gfd) after 95 hours of operation and 0.81 m3/m2/d (19.8 gfd) after 238 hours.
This run did not last long enough to determine the upper limit on solids concentration in the con-
centrated feed, but excellent rejections of solids and color were obtained (see Table 14). Run H-2
was begun after the membrane had been left dry for 2 weeks, and the fluxes and rejections were not
as high overall as in Run H-1. The initial flux was higher than Run H-1 (3.30 m3/m2/d [81.0 gfd])
but dropped to 1.53 m3/m2/d (37.6 gfd) after only 18 hours. At 97 hours of operation the flux
was only 0.46 m3/m2/d (11.3 gfd) compared with the value (1.31 m3/m2/d [32.2 gfd]) for Run
H-1. The flux for this membrane dropped more rapidly and leveled off at a lower value than Run
H-1, indicating that the drying out of the polyacrylic acid layer had adversely affected performance.
As seen in Table 14, the rejections of the various constituents were lower also. Run H-3 was started
with a membrane that was apparently affected by soluble iron in the membrane formation step. The
initial flux was 4.12 m3/m2/d (101.1 gfd) with contaminated permeate and 3.72 m3/m2/d 91.3 gfd)
after the permeate had cleared. This membrane was lost because of breakage and was replaced with
a second membrane that gave an initial flux of 3.95 m3/m2/d (67.0 gfd) within 17 hours that
dropped to 1.40 m3/m2/d (34.4 gfd) in 67 hours. The rejections during this run varied somewhat
but were generally high, as seen in Table 14.
The effect of a wash sequence upon membrane performance was seen In two instances. In
Run H-2, a NajCOa (pH = 10.6) solution that was used to wash the membrane raised the flux from
1.31 m3/m2/d (32.2 gfd) to 1.80 m3/m2/d (44.2 gfd). Unfortunately, the tube bundles were
broken before further operation could be started. Another attempt at washing the membrane was
made during Run H-3, using a NaOH (pH = 11.9) solution. The flux was raised to 1.74 m3/m2/d
(42.8 gfd), but it dropped to 1.09 m3/m2/d (26.7 gfd) within 1 day of operation. Unfortunately,
the unit was shut down before the effect of the wash on the rejections could be observed.
83
-------
TABLE 14. SUMMARY OF REVERSE OSMOSIS DATA
% Rejection of
Conductivity
Total solids
Ash
Volatile solids
Color
Chlorides
Sodium
BODS
COD
TOC
Total sulfur
Run H-1
81.0-95.3
97.8 - 99.6
96.5 - 99.1
99.4-100.0
99.9-100.0
79.2 - 92.9
N/A
N/A
N/A
N/A
N/A
Run H-2
76.9 - 88.2
91.2-95.0
89.9 - 96.5
94.3 - 92.3
97.8 - 99.8
N/A
N/A
N/A
N/A
N/A
N/A
Run H-3
51.5-98.3
88.7 - 97.2
86.6 - 98.9
87.4 - 97.1
98.2-100.0
84.5-91.7
82.9 - 99.7
72.4 - 90.6
60.3 - 95.2
25.4-95.1
90.1 - 98.6
Note: N/A = Not available.
It is evident from the problems encountered during this study that the membrane formation
step has not been optimized. In each run, more membrane forming material was used during the
laboratory trials than should have been necessary. The reason for this is unknown, but it is possi-
bly attributable to false fluxes during the formation step because of the weak spots in the mem-
brane at points where the tubes rub each other and/or the piping. The higher apparent flux caused
by these spots could falsely indicate that more membrane material is needed, while the bulk of the
tube surface is coated too heavily. This could adversely affect the flux rate. Another interesting
phenomenon observed during this study was that the permeate flow is initially dark and contami-
nated on startup of the unit; this occurred just after new membrane formation as well as after an
intermittent shutdown. The permeate quality gradually improved until a very clear solution was
obtained, usually after approximately 20 minutes. The flux at startup is also slightly higher before
the permeate clears. The reason for this phenomenon is unknown but is possibly caused by com-
paction of the polyacrylic acid layer or by the formation of a wood-chemical type membrane. This
could also be connected with the sudden drop in flux in Run H-2, which resulted in the membrane
being dried out. The constituents in the waste water could have filled in membrane defects, causing
a drop in flux. It must be emphasized that these theories have not been confirmed because of the
operational problems encountered.
During the course of these trials, several points relative to the design of dynamic membrane
reverse osmosis units for treatment of pulp and paper mill effluents were realized. The most ob-
vious is that adequate prefiltration of the feed must be provided to prevent fiber accumulation from
shutting the units down. The outside-flow type tube bundle could only be practical when such pre-
treatment is used, because even if the tubes did not break, the flow through a clogged bundle would
be drastically altered. The use of an inside-flow tubular membrane support would be advisable if
the porous tube material is suitable for dynamic membrane formation and can withstand the high
internal pressure. This configuration would lessen the probability of fibers and solids accumulating
on membrane surfaces, and a smooth continuous membrane could be formed more easily. The use
of a pH control system was found to be unnecessary because the waste water pH did not go much
higher than 10.0.
The rejections of the various parameters indicate that this type of membrane system could be
developed to concentrate decker filtrates, or other waste waters containing pulping chemicals, to
produce solutions concentrated enough for economical recovery of chemicals. The large organic
molecules which contribute to color were removed almost entirely, and sodium and total sulfur
84
-------
rejections were also high (see Table 6). The chlorides were rejected well, but the actual chloride
concentration was low and not a major concern. The BODS, COD, and TOC rejections indicate
that the permeate should be useable for recycle purposes, especially because the unit can treat hot
waste waters.
ULTRAFILTRATION UNIT
As with the reverse osmosis unit, equipment malfunctions and processes with the membrane
support system (piping and tube bundles) prevented the accumulation of data which could be
reliably used for scale-up purposes. However, the pilot operations that were performed provided
information which will be invaluable for future work with dynamic membranes. The data from the
six ultrafiltration runs are summarized in Table 15 and are indicative of the differences in operation.
Run U-1 was the best with regard to rejection of solutes at high flux rates. Runs U-2 and U-3 had
good color rejections but poor solids rejections and also poor flux rates. Run U-4, the first trial with
the carbon tubes, obtained better fluxes with fairly good color removals but only fair total solids
rejection. The fair to good rejections of TOC, COD, and BODS indicated, in conjunction with the
previous statement, that the unit was rejecting the large organic molecules which comprise the color
bodies, while the smaller organics were allowed to pass through. Runs U-5 and U-6 started poorly,
but membrane washes and extra layers of membrane improved both fluxes and rejections to a level
approaching those of Run U-1. In the last day of Run U-5, the fluxes and rejections from individual
ceramic tube bundles were determined. The data for these bundles are given in Table 16. The tube
bundles did not show any distinct trends regarding flux vs. rejection, nor did any trends become evi-
dent in the position of a single bundle in the series. This randomness and the previous problems with
the bundles in this run indicate that the problems in membrane formation, operation, and clean up are
related to individual bundles rather than the system as a whole. This tends to reinforce the concept
that dynamic membranes are a viable system and that the problems encountered here were peculiar
to this particular unit. If the poor fluxes and rejections were a membrane problem rather than a
mechanical problem, one would expect uniformly poor individual performances. If tubes were
rubbing each other or the piping, causing uneven membrane formations, one would expect uneven
fluxes.
The overall performance when a successful membrane formation (such as that of Run U-1) is
completed is very good, with high fluxes and rejection. The membrane will probably need a water
rinsing about every 2 weeks, depending on the solids level maintained in the concentrate stream.
It is possible to achieve a total solids level of over 10% with this type of system and to then intro-
duce this stream into the mill recovery system. A high level of organics removal with low chloride
buildup is possible; this is a desirable feature when treating bleach plant liquors. The color removal
is usually quite good even with poor membrane performance.
In addition to these general observations, there were several specific items worthy of notice. The
permeate was always dark in color when the unit was first started, whether with a new membrane or
after an intermittent stoppage. The permeate would gradually clear up in about 15 to 30 minutes.
In Runs U-1, U-5, and U-6, the rejections of total solids, ash, COD, and TOC were always lower dur-
ing the first day of a run than later days. This tended to support the theory that a lignin-type mem-
brane, formed from wood chemical constituents in the feed, was being formed over the hydrous
Zr(IV)-SHlV) oxide membrane.
85
-------
TABLE 15. SUMMARY OF ULTRAFILTRATION DATA
% Rejection of
Conductivity
Total solids
Ash
Volatile solids
Color
Chlorides
Sodium
BOD5
COD
TOC
Total sulfur
Run U-1
26.0-67.4
56.0-97.7
42.1-93.6
67.2-98.7
98.9-99.7
5.5-23.8
N/A
N/A
N/A
N/A
N/A
Run U-2
24.0-42.9
64.0-86.9
39.3-76.6
81.7-92.3
94.3-99.1
N/A
N/A
N/A
N/A
N/A
N/A
Run U-3
23.0-24.0
55.3-57.0
6.5-19.3
72.8-73.1
94.2-95.8
N/A
N/A
N/A
N/A
N/A
N/A
Run U-4
19.7-31.1
52.5-58.6
30.7-41.4
61.3-85.3
39.4-98.5
10.0-33.0
31.0-61.7
76.2
71.1-87.6
84.4-92.6
37.5-79.4
Run U-5
17.9-44.4
50.9-94.3
31.9-83.5
73.3-96.9
96.7-98.8
1.1-39.3
3.8-84.6
72.8-92.8
82.2-97.7
81.8-97.4
25.8-92.9
Run U-6
10.3-35.6
45.6-93.6
27.9-80.5
69.4-96.7
96.3-99.3
4.4-41.7
31.9-82.0
74.1-92.6
58.8-97.5
82.8-97.1
48.3-92.9
Note: N/A = Not Available.
TABLE 16. ULTRAFILTRATION DATA, INDIVIDUAL BUNDLES,
RUN U-5, DECEMBER 7, 1977
Flux,
Bundle1
1
2
4
5
6
8
9
10
11
12
13
14
15
16
Total combined
m3/m2/d
3.15
5.48
2.98
4.72
4.14
4.25
3.95
3.62
2.98
6.29
4.25
1.89
3.78
3.47
4.05
(gfd)
( 77.3)
(134.6)
( 73.2)
(115.9)
(101.7)
(104.3)
( 97.0)
( 88.8)
< 73.2)
(154.5)
(104.3)
( 46.4)
( 92.7)
( 85.1)
( 99.3)
Total
Solids
95.6
96.1
94.3
94.5
95.6
95.2
94.1
94.6
94.4
94.9
93.2
93.8
94.8
93.2
94.2
Ash
89.0
88.0
88.2
85.2
88.1
88.1
85.9
86.5
87.5
85.0
81.0
83.2
86.8
77.4
83.8
Volatile
Solids
97.3
98.2
96.5
97.0
97.6
97.1
96.2
96.7
96.1
97.5
96.4
96.5
96.9
97.3
96.9
Color
98.4
99.6
98.4
98.6
99.1
98.8
98.1
98.8
98.6
99.4
97.6
97.4
98.4
97.3
98.4
Cent Reje
BODS
93.1
93.9
92.2
92.2
93.3
91.9
90.8
91.1
90.3
91.9
86.9
87.2
91.7
90.8
90.0
COD
96.6
98.9
97.8
97.7
98.5
98.2
97.8
98.1
97.8
98.9
97.4
97.5
97.9
97.3
97.7
TOC
97.6
98.7
97.2
97.4
97.9
97.3
96.8
97.5
97.2
98.0
96.6
96.6
97.4
96.3
96.8
Sodium
90.0
88.3
85.1
85.2
87.9
87.4
86.0
85.3
85.1
85.4
83.0
84.9
86.7
83.6
84.6
Total
Sulfur
5.16
5.16
93.6
N/A
12.9
N/A
14.8
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Bundle number indicates position in series sequence of bundles.
Note: N/A = Not Available.
The effect of washing the membrane was noted during several runs. At the end of Run U-1, a
water rinse was performed, and this improved the flux from 2.25 m3/m2/d (55.2 gfd) to 6.36
m3/m2/d (156.1 gfd). Mechanical problems precluded evaluation of the long-term effect of this
wash. The tubes in Run U-2 were washed three separate times, but no significant improvement
was obtained. The same result was noted in Runs U-3 and U-4, with the conclusion being that a
washing sequence will not aid a poor membrane but will restore performance to a good one. When
simple washes failed to improve fluxes and rejections, another layer of membrane would sometimes
help. This was observed in Runs U-5 and U-6, in which two separate membrane additions improved
performance dramatically. This emphasizes a useful feature of dynamic membrane systems in that
a membrane can be adjusted if the initial performance is not satisfactory.
An interesting observation is that the membrane used in Run U-1 was 4 months old when the
run began. The membrane had been stored by simply filling the unit with deionized water. This
indicates that dynamic membrane modules can be generated separately from the ultrafiltration
86
-------
unit and stored until needed. This would allow a great degree of freedom in operation because a
faulty module (if the system is so designed) could be isolated, removed, and replaced; and the
membrane could be washed and regenerated by the user at his convenience using a small scale unit.
The flux rate in Run U-1 held steady while the total solids concentration in the concentrate
stream rose. The flux started to drop slightly when the solids concentration reached 8% or higher,
indicating that operation at high solids levels is possible. The pressure drop through the unit leveled
off and held steady throughout the run. This conflicted with the usual expectation that the pres-
sure drop would rise in proportion to the increase in solids.
The overall flux rate rose as the temperature got higher until the set point of the temperature
control system was reached. This emphasizes an advantage of dynamic membrane systems in that
higher temperatures cannot only be tolerated but can produce higher flux rates.
If tube bundles are used with an outside flow arrangement, prefiltration of the feed to remove
fibers is absolutely necessary. Fiber clogging the bundles cannot be tolerated. No pH control was
needed because the pH of the bleach plant waste waters rarely rose above 10.0.
WATER RECOVERY
In any discussion of the capabilities of a reverse osmosis or ultrafi It ration unit, the percent
water recovery achieved by the unit is an important consideration. The percent water recovery is
defined as:
% Ru O = ( volume of permeate produced/unit time \ -JQQ |QJ
Vvolume of concentrate introduced/unit time/
and represents a measure of how much of the water fed to the unit is recovered. The use of this
parameter is based on a constant membrane surface area and a controlled flow regime to achieve
the desired percent recovery. Unfortunately, this was not the case with the units in this pilot
operation. The flow was controlled to a rate which would produce the desired flow velocity across
the tubes, and it was not adjusted to produce a controlled water recovery as the membrane surface
area changed because of tube bundle breakage. The tube bundles broke so frequently that a con-
stant surface area was not possible throughout any particular run, and no reasonable comparison of
water recoveries from run-to-run or day-to-day was possible. The runs which gave relatively stable
and high fluxes were not comparable with laboratory data because the surface areas changed dras-
tically during the runs. Because flux was reported in terms of volume per unit surface area per day,
no indication of the changes in total surface area was apparent. For example, Run H-2 started with
29 bundles and was terminated when only 3 bundles remained in operating condition. Thus, no
attempt at presenting water recovery data was made in this report.
ECONOMIC EVALUATION
The primary purpose of this pilot plant project was to develop operating and scale-up data
which could be used to design a commercial scale installation for the treatment of pulp mill waste
waters. This includes defining certain operational parameters of the units, such as the expected
87
-------
fluxes, membrane life, effectiveness of cleaning procedures, equipment reliability, and long-term
operating efficiency of the units. The pilot plant study described in this report failed to define these
parameters adequately, and in some instances it stressed that the units needed redesigning. A steady-
state flux was not achieved in either unit, although it appears that 1.63 m3/m2/d (40 gfd) for the
reverse osmosis (hyperfiltration) unit and approximately 4.48 mVmVd (110 gfd) for the ultra-
filtration unit would be about optimum. Membrane life was not defined in either case, because
neither unit operated long enough to test this parameter. Some tests of cleaning the membrane
were made, but not enough data were available to give any clear indication of cleanability, except
that a good membrane could be cleaned significantly by a Na2C03 solution. The reliability of the
equipment was poor, although it is important to note that the pilot unit was not a commercial unit
and definitely is not indicative of the present state-of-the-art in tubular dynamic membrane units.
The data clearly show that no long-term reliability or efficiency can be estimated. Thus, any
economic evaluation based on this pilot operation would be speculative, and any scale-up thus
derived could be grossly undersized or oversized. Previous studies (26, 41) with textile waste waters
have been more successful and have indicated that dynamic membrane"systems are competitive with
other membrane systems. The dynamic membrane system is being studied further for use with pulp
mill waste waters.
-------
REFERENCES
1. Reid, C. E. and E. J. Breton. Water and Ion Flow Across Cellulosic Membranes. Journal of
Applied Polymer Science I (2): 133-43, 1959.
2. Kesting, R. E. and J. Eberlin. Semipermeable Membranes of Cellulose Acetate for Desalination
in the Process of Reverse Osmosis. IV. Transport Phenomena Involving Aqueous Solutions
of Organic Compounds. J. Applied Polymer Science U): 961 - 967, 1966.
3. Matsuura, T. and S. Sourirajan. Physicochemical Criteria for Reverse Osmosis Separation of
Alcohols, Phenols and Monocarboxylic Acids in Aqueous Solutions Using Porous Cellulose
Acetate Membranes. J. Applied Polymer Science ]j>: 2905 - 2927, 1971.
4. ' Sourirajan, S. and T. Matsuura. Transportation Through Reverse Osmosis Membranes.
Reverse Osmosis and Synthetic Membranes, TheoryTechnologyEngineering,
S. Sourirajan, ed. Report No. NRCC No. 15627, National Research Council Canada, Ottawa,
Canada, 1977, Chapters.
5. Glueckauf, E., and D. C. Sammon. Transport of Ions and Water Through Cellulose Acetate
Membranes. 3rd International Symposium on Fresh Water from the Sea, 2, 1970. pp. 397-422.
6. Wiley, A. J., Dubey, G. A., and I. K. Bansal. Reverse Osmosis Concentration of Dilute Pulp
and Paper Effluents. EPA Project 12040 EEL, U. S. Environmental Protection Agency,
Cincinnati, Ohio, 1972.
7. Duvel, W. A., Jr., and T. Helfgott. Removal of Wastewater Organics by Reverse Osmosis.
J. Water Pollution Control Federation,£7(1): 57, 1975.
8 Chian, E. S. K. and H. H. P. Fang. Physicochemical Criteria of Removal of Trace Organics by
Reverse Osmosis. In: Water-1976. I. Physical, Chemical Wastewater Treatment, A.I.Ch.E.
Symp. Series 73(166): 152 - 161, 1977.
9. Fang, H. H. P. and E. S. K. Chian. Reverse Osmosis Separation of Polar Organic Compounds
in Aqueous Solution. Environmental Science and Technology ^0(4): 364 369, 1976.
10. Nelson, W. R., Walraven, G. 0., and D. C. Morris. NSSC Mill Experience with Waste Water
Reuse and Reverse Osmosis. In: Proceedings of the TAPPI Environmental Conference, New
Orleans, La., 1974, pp. 63 - 72.
11. Matsuura, T. and S. Sourirajan. Physicochemical Criteria for Reverse Osmosis Separation of
Aldehydes, Ketones, Ethers, Esters, and Amines in Aqueous Solutions Using Porous Cellulose
Acetate Membranes. J. Applied Polymer Science Jfr 1663 - 1686, 1972.
89
-------
12. Fang, H. H. P. and E. S. K. Chian. Removal of Alcohols, Amines, and Aliphatic Acids in
Aqueous Solution by NS - 100 Membrane. J. Applied Polymer Science 19: 1347 - 1358,
1975.
13. Crowley's Milk Company. Membrane Processing of Cottage Cheese Whey for Pollution
Abatement. EPA Report No. 12060 DXF 07/71, U. S. Environmental Protection Agency,
Cincinnati, Ohio, 1971.
14. Kaup, E. C. Design Factors in Reverse Osmosis. Chemical Engineering, 80(8): 46 - 55,
April 2, 1973.
15. Sourirajan, S. Reverse Osmosis,p. 261. Academic Press, New York, New York, 1972. 580 pp.
16. Weber, W. J., Jr. Physicochemical Processes for Water Quality Control, p. 322. Wiley
Interscience. New York, New York, 1972. 640pp.
17. Sourirajan, S. Reverse OsmosisAccomplishments and Prospects. Reverse Osmosis and
Synthetic Membranes, TheoryTechnologyEngineering. S. Sourirajan, ed. Report No. NRCC
No. 15627, National Research Council Canada, Ottawa, Canada, 1977, Chapter 28.
18. De Danske Sukkerfabrikker. DOS Reverse Osmosis System Technical Bulletin. De Danske
Sukkerfabrikker, DK-4900, Nakskov, Denmark, 1977.
19. Fremont, H. A., Tate, D. C., and R. L. Goldsmith. Color Removal from Kraft Mill Effluents
by Ultrafiltration. EPA 660/273-019, U. S. Environmental Protection Agency, Cincinnati,
Ohio, 1973. 240pp.
20. Quinn, R. K., and W. K. Hendershaw. A Comparison of Current Membrane Systems Used in
Ultrafiltration and Reverse Osmosis. Industrial Water Engineering, June/July, 1976, p. 12.
21. Sliger, H. B., and R. Quinn. Application of Membrane Processes. Desalination, 19:
573-586,1976. ~~
22. Bansal, I. K. Concentration of Oily and Latex Waste Waters Using Ultrafiltration Inorganic
Membranes. Industrial Water Engineering, Oct./Nov. 1976, pp. 6 11.
23. Bansal, I. K. Reverse Osmosis and Ultrafiltration of Oily and Pulping Effluents. Industrial
Wastes, May/June, 1977, p. 32.
24. Morris, D. C., Nelson, W. R., and G. O. Walraven. Recycle of PaperMill Waste Waters and
Application of Reverse Osmosis. Report No. 12040 FUB 01/72, U. S. Environmental
Protection Agency, Cincinnati, Ohio, 1972.
25. Rozelle, L. T., Kopp, C. V., Jr., and K. E. Cobian. New Membranes for Reverse Osmosis
Treatment of Metal Finishing Effluents. EPA 660/2-73-033, U. S. Environmental
Protection Agency, Cincinnati, Ohio, 1973.
90
-------
26. Brandon, C. A., and J. J. Porter. Hyperfiltration for Renovation of Textile Finishing Plant
Waste Water. EPA 600/2-76-060, U. S. Environmental Protection Agency, Cincinnati, Ohio,
1976.
27. Donnelly, R. G., Goldsmith, R. L, McNulty, K. J., Grant, D. C., and M. Tan. Treatment of
Electroplating Wastes by Reverse Osmosis. EPA 600/2-76-261, U. S. Environmental
Protection Agency, Cincinnati, Ohio, 1976.
28. Porter, J. J., and J. L. Edwards, Jr. A Study of Comparisons of Three Membrane Systems
Relative to Discharge Water from Kraft Pulp and Paper Mill Process Streams. Southern Pulp
and Paper Manufacturer, 40(12): 24-31, 1977.
29. Luttinger, L. B., and G. Hoche1. Reverse Osmosis Treatment with Predictable Water Quality.
Environmental Science and Technology, JH7): 614 618, 1974.
30. Nusbaum, I., Cruver, R. E., and J. H. Sleigh, Jr. Reverse Osmosis New Solutions and New
Problems. Chemical Engineering Progress, 68(1): 69 - 70, 1972.
31. Fang, H. H. P., and E. S. K. Chian. Removal of Dissolved Solids by Reverse Osmosis. In:
Water - 1976. I. Physical, Chemical Wastewater Treatment, A.I.Ch.E. Symp. Series ^73(166):
137- 143, 1977.
32. Bansal, I. K. Progress in Developing Membrane Systems for Treatment of Forest-Products and
Food Processing Effluents. In: Water 1976. I. Physical, Chemical Waste Water Treatment,
AJ.Ch.E. Symp. Series 73(166): 144-151,1977.
33. Maples, G., and E. W. Lang. Studies of Membrane Processes for Pulp Mill Pollution Control.
In: Proceedings of theTAPPI Environmental Conference, Washington, D. C., April 1978,
pp. 71 - 82.
34. Riley, R. .L., Fox, R. L, Lyons, C. R., Milstead, C. E., Seroy, M. W., and M. Tagami. Spiral
Wound Poly (Ether/Amide) Thin Film Composite Membrane Systems. In: Proceeding of
the Membrane Separation Technology Conference, Clemson University, Clemson, S. Car.,
1976.
35. Allegrezza, A. E. Jr., Carpentier, J. M., Davis, R. B., and M. J. Coplan. Hollow Fiber Reverse
Osmosis Membranes. In: 1976 Water. I. Physical, Chemical Waste Water Treatment,
A.I.Ch.E. Symp. Series 73(166): 162 - 165, 1977.
36. Thomas, D. G. Dynamic Membranes Their Technological and Engineering Aspects.
Chapter 14 of Reverse Osmosis and Synthetic Membranes, Theory Technology
Engineering. S. Sourirajan, ed. Report No. NRCC No. 15627, National Research Council
Canada, Ottawa, Canada, 1977. pp. 295 - 312.
37. Kraus, K. A., Shor, A. J., and Johnson, J. S., Jr. Hyperfiltration Studies X. Hyperfiltration
with Dynamically Formed Membranes. Desalination, 2: 243 266, 1967.
91
-------
38. Bansal, I. K., and A. J. Wiley. Improving Reverse Osmosis Performance with Dynamically-
Formed Wood Chemical Membranes. In: Proceedings of the TAPPI Environmental
Conference, San Francisco, CA, 1973. pp. 245 - 251.
39. Perona, J. J., Butt, F. H., Fleming, S. M., Mayr, S. T., Spitz, R. A., Brown, M. K., Cochran,
H. D., Kraus, K. A., and J. S. Johnson, Jr. Hyperfiltration. Processing of Pulp Mill Sulfite
Wastes with a Membrane Dynamically Formed from Feed Constituents. Environmental
Science and Technology, 1_(12): 991 - 996, 1967.
40. Johnson, J. S., Jr., Minturn, R. E., and G. E. Moore. Filtration Techniques for Purification
of Kraft Mill and Bleach Plant Wastes. TAPPI 57(1): 134 - 138, 1974.
41. Brandon, C. A., Johnson, J. S., Jr., Minturn, R. E., and J. J. Porter. Complete Reuse of
Textile Dyeing Wastes Processed with Dynamic Membrane Hyperfiltration. Textile Chemist
and Colorist, 5(7): 134 - 137, 1973.
42. Johnson, J. S., Jr. Advanced Filtration of Waste Waters. In: Chemical Division Annual
Report, Oak Ridge National Laboratory, Period Endmg May 20, 1974.
92
-------
APPENDIX A
PRINCIPLES OF OPERATION
Osmosis is the natural process which occurs when two solutions of unequal concentrations of
solute(s) are separated by a semipermeable membrane. The process depends upon the existence of
such a membrane, which is capable of allowing the passage of the solvent molecules while retaining
the solute molecules. The driving force behind the process of osmosis is the vapor pressure
differential between the dilute and concentrated solutions. Thus, passage of solvent molecules
through the membrane into the concentrated solution will occur until thermodynamic equilibrium
is reached or until an external force interferes before the system reaches the lowest chemical potential.
The osmotic pressure (IT) is the pressure which must be applied to the concentrated solution to
prevent passage of the solvent molecules through the semipermeable membrane. The osmotic
pressure of a solution is dependent upon the solute concentration, activity coefficient, the degree of
ionization of the solutes, and the temperature. It is defined as
TT = iRTC (1)
where ir = osmotic pressure, atm
i = van't Hoffs factor, which is dependent upon the activity coefficient and the degree
of ionization
R = gas constant = 82.1 atm cm3/g-mole °K
T = temperature, °K
C = concentration of solute, g mole/cm3
For an ideal dilute solution, i is a constant, so that the osmotic pressure (IT) is directly proportional
to temperature and concentration. For pure water, IT is equal to zero (6).
Reverse osmosis is the reversing of the natural process by applying an external driving force to
overcome the existing osmotic pressure. If a pressure (P) which is greater than TT, is applied to the
more concentrated solution, the solvent will flow from that side through the membrane into the
dilute solution. For any given membrane, the amount of water flowing opposite to the osmotic
flow is directly proportional to the differential applied pressure. This flow is measured in terms
of volume per unit surface area of membrane and is defined as
J = A(Ap-ATr) (2)
where J = flux rate through the membrane, gfd
93
-------
A = water permeability coefficient, gfd/psig
Ap = difference between the applied pressure and the delivery pressure of product
water, psig
ATT = (difference between the liquor and product osmotic pressures) + (osmotic
pressure increase because of fouling and concentration polarization), psig
The product water is delivered at atmospheric pressure, and the osmotic pressure of the
product water is very small compared with that of the liquor. Assuming no fouling and concentra-
tion polarization effects, the driving force (Ap - ATT) becomes equal to the difference between the
applied driving force, PA, and the osmotic pressure of the liquor (IT). Then Equation 2 becomes
J = A(PA-;r) (3)
As seen in the Equation 3, as the liquor becomes more concentrated, the osmotic pressure
increases, and higher applied pressures are necessary to maintain the flux rate, J.
Although both reverse osmosis and ultrafiltration are processes which involve transport of
solute molecules through a semipermeable membrane by applying pressure, the mechanisms,
capabilities, and results from the two are quite different. The main differences are the membranes
used (and the subsequent mechanism involved), the operating pressure, the size and types of
solutes which can be removed, and the quality of the permeates. The membranes used for ultra-
filtration are "loose" membranes, meaning that they retain only the higher molecular weight
organic molecules. The pore diameters are in the range of 2x10~8 to 1x10~s meters (200 to
100,000 angstroms). The membranes used for reverse osmosis are "tighter" and retain low
molecular weight organic molecules and inorganic salts. The reverse osmosis membranes are
generally asymmetrical, with a macroporous substructure and a dense surface layer. The
mechanisms of removal in both reverse osmosis and ultrafiltration are similar in that they both in-
volve a rejection of solute molecules by a semipermeable membrane under the driving force of
applied pressure. However, the ultrafiltration process, which operates in a low pressure range of
3.5x10s to 2x106 Pa(50-300 psi), is a simple sieving operation, separating the solute molecules
because of their size or shape. Reverse osmosis, which operates in a pressure range of 1.4x106 to
1x107 Pa(200-1500 psi), removes solutes mainly by surface interaction between membrane and
solute, and to a lesser degree, by physical sieving.
The mechanism involved in reverse osmosis is not fully understood, and two main schools of
thought exist. The solution-diffusion theories are based on the concept of the water and any solute
molecules which are capable of hydrogen bonding being bonded to available hydrogen bonding sites
within the membrane and subsequent transport occurring because of diffusion. Reid and Breton,
(1) and Kesting (2) are proponents of this theory but have differing views on the function of
the pores in the membrane. Reid and Breton proposed that nonhydrogen bonding species of small
enough diameter pass through the pores when the pores are not blocked wi|h bound water,
whereas Kesting advocated the view that the pores are membrane imperfections with passage of
solutes being size selective (2). The most widely accepted theory is the preferential sorption-
capillary flow mechanism proposed by Sourirajan (3, 4). In this theory, the process is one of
adsorbing the solution components onto the membrane surface, and the subsequent transport is a
capillary flow phenomenon through the pores. Separation of solvent and solute depends on the
94
-------
chemical affinity of the membrane for the solutes, and the transport rate depends upon the size and
number of pores and the applied driving force (5). Another pore theory, proposed by Glueckauf,
states that the adsorption is less important than the ionic repulsion which takes place within the
pores of a membrane with a low dielectric constant material (5). No one theory is universally
accepted because each has its faults, but the Sourirajan preferential sorption-capillary flow seems to
be the most widely accepted.
The performance of ultrafiltration and reverse osmosis units is usually defined by the percent
rejection of the solute in question and the flux rate through the membrane. The percent rejection
is defined by
{' CP\
%R=(1_J. x 100 (4)
where % R = percent rejection
Cp = concentration of the permeate
Cp = concentration of the feed to the module
The % R is affected by many factors, such as the polarity of the membrane, steric effects, ioniza-
tion and hydration of the solutes, pH, applied pressure hydrogen bonding, and the valence (if
removing electrolytes). The more nonpolar a membrane is the better the rejection, because there
is less chemical affinity for the membrane by the solute (8). Steric effects are caused by the resis-
tance to transport through the pores because of the size or shape of the molecule to be removed. High
molecular weight organic molecules are often too large to fit through the pores. Compounds with
the same molecular weight but different configurations are often removed to different degrees.
For example, a branched alcohol is removed more completely than a straight chain alcohol, and
secondary amines are removed more completely than primary amines. The pH and degree of
ionization together can radically affect removals. At a pH where the molecule is only partially
dissociated, removals can be markedly lower than if the molecule is completely dissociated (3, 7, 8,
10). Usually there is a linear improvement in rejection with the degree of dissociation (8, 9). The
separation of formic acid has been reported at 6% when partially dissociated and up to 98% when
fully dissociated (9). The importance of the hydrogen bonding ability of the solute has been
recognized, in that molecules which can form hydrogen bonds permeate through the membranes
more readily than those which cannot, giving lower rejections (2, 3, 7, 1 1). The enthalpy of
hydration of inorganic ions is recognized as a factor in rejection. The higher the enthalpy of
hydration, the bulkier the ion-bound water complex, and the steric effects again become important
(9, 12). It has also been found that the rejection of electrolytes increases with valence, with bi-
valent and trivalent ions being rejected more than monovalent ions. The rejection is also known to
improve with increasing pressure, although this effect is more pronounced with solutes which are
otherwise poorly removed (8, 13). The main restraints on applied pressure are the mechanical
strength of the membrane supports and the adverse effect of membrane compaction, which tends
to reduce the flux rates (13).
The ffux rate through the membrane is critical to the design of the equipment because it
determines the total surface area of membrane necessary for efficient operation. The flux rates
for ultrafiltration units are fairty high, usually ranging from 2 to 6 m3/m2/d (50 to 150 gfd).
95
-------
Reverse osmosis units tend to have low flux rates, usually ranging from 0.4 to 1 mVmVd (10 to 25
gfd). Several of the factors which affect rejection also affect the flux rates. The polarity of the
membrane inversely affects flux compared with rejection in that the-more nonpolar the membrane,
the lower the flux. As shown in Equation 3, flux rates should increase with higher pressures up to
the previously mentioned membrane compaction point (13, 14). It is possible to attain higher flux
rates by raising the operating temperature, but this is limited by the type of membrane used. The
most important factors in maintaining high flux rates are prevention of fouling and concentration
polarization. Fouling is a phenomenon which is not well defined but is unique in each situation.
Essentially, in many waste-water treatment operations there are constituents in the feed which
adversely affect performance by adhering to the membrane surface and lowering the flux rates and
in some cases the rejections.
Concentration polarization is a phenomenon which occurs when the concentration of solute
on the feed side membrane surface increases to the point that a concentration gradient between the
membrane and the bulk solution develops (4, 6, 14, 15, 16). This causes a flux decline because the
diffusion of solute away from the membrane surface cannot remove the buildup of a boundary
layer. The concentration polarization ratio is defined as (14)
CM (5)
CP=
CB
where CP = concentration polarization ratio
CM = solute concentration at the membrane
Cg = bulk solute concentration
As the boundary layer builds up, the ratio increases and the osmotic pressure is higher. As a
result, the driving force is lowered, unless the applied pressure (P^) is increased proportionately.
For example, a unit operating at 6.89x106 Pa (1000 psi) treating a solution containing 35,000 mg/1
total dissolved solids (CP -1.2, osmotic pressure 2.4x106 Pa[350 psi]) has a boundary concentra-
tion of 42,000 mg/1 total dissolved solids, with an osmotic pressure of 2.9x106 Pa(420 psi). If the
feed rate is slowed down to obtain higher water recovery so that the boundary concentration
increases 100%, the osmotic pressure is then 4.8x106 (700 psi). Thus, instead of a driving force of
1000 - 420 = 580 psi, it is now 1000 - 700 = 300 psi (13).
Another type of fouling occurs when the boundary layer becomes supersaturated with soluble
inorganic salts, which precipitate out of solution and plug the membrane. The result is a loss of flux
and possible damage to the membrane (14).
It is often possible to avoid the effects of fouling, concentration polarization, and scaling by
altering the method of operation and/or redesigning the modules which house the membranes.
Fouling can be avoided or lessened by pretreating the feed to remove the fouling components or
providing enough velocity to flush fouling layers from the membranes. Separate washings and
"pausing" techniques which allow the natural osmosis process to flush out the membrane are also
helpful. To eliminate concentration polarization, it is necessary to prevent the formation of the
boundary layer by using high velocity flow over the membrane, or by designing the modules to
produce adequate turbulence. Scaling can be prevented by pretreatment, pH adjustment, or by
keeping the product-to-feed ratio low to ensure that the supersaturated layer does not form (19).
96
-------
APPENDIX B
MEMBRANE SUPPORTS
In the design and operation of reverse osmosis and ultrafiltration units, the two most
important choices to be made are those of the proper membrane material and the most efficient
membrane support system. There are four main types of membrane supports: plate and frame,
spiral wound, hollow fiber, and tubular. Each has its variations, relative advantages, and disadvan-
tages.
Plate and Frame
These modules consist of flat membranes supported by porous plates and assembled in a stack
of closely spaced plates. The feed flows over the membrane and the permeate is drawn off through
the porous plates. The permeates can be collected with either a central tube running through the
centers of the plates, or with individual tubes at the edge of each plate (17, 18). The plates are
closely spaced to prevent laminar flow, and turbulence promoters are used to help control fouling.
Spiral Wound
The spiral-wound module is a unit in which two flat membranes, separated by a porous
support, are wrapped around a porous collection pipe like a jelly roll. The edges of the two
membranes are sealed together, and a spacer is provided between the membrane surfaces to allow
axial flow of the feed. The permeate flows through the porous support to the central collection
pipe (19,20).
Hollow Fiber
This system consists of thousands of very fine hollow fibers made of the membrane
material itself. The earliest models used external feed flow, in which a bundle of hollow
fibers was folded double and the open ends set in a bonding resinous material. The feed flowed
under pressure over the outside of these fibers, and the permeate passed out the ends of the fibers
for collection. More recently, fibers have been developed which allow internal feed flow through
the fibers. This has an advantage over the outside flow as this configuration is less likely to foul and
easier to backwash (20, 21).
Tubular
This module consists of a membrane applied either on the inside or outside of a
porous tube. The feed flows tangientially over the membrane and the permeate is collected on the
other side of the tube wall. Various materials have been used for the porous tubes: spiral-wound
fiberglass, carbon, stainless steel, ceramic, and thin polymeric tubes encased in sand (for strength)
(6, 22, 23, 24, 25).
Comparison
Several studies have been performed in which two or more of these module types were operated
97
-------
and compared (2, 6, 9, 21, 24, 26, 27, 28). Each configuration has its own intrinsic advantages and
disadvantages. The proper choice of a membrane support is a balance of efficiency and economy.
The plate and frame configuration was the first to be used in large scale reverse osmosis appli-
cations. The system requires large equipment, high maintenance costs, and a large space for the
equipment installation. This module is more prone to foul than others, and damaged membranes
are more difficult to replace than with the other units. These units are presently being marketed in
an improved version, but use of plate and frame systems is not yet widespread (17,18).
The spiral-wound module has a large total membrane surface area per unit volume of equip-
ment, good production efficiency, and low cost per gallon of water (permeate) produced. The early
models had problems with fouling, plugging, bad flow distribution, and poor overall reliability (19).
The recently developed modules, spacers, turbulence promoters, and new designs have improved
performance, and the fouling and plugging problems are minimal and cleaning is easier. Even so, if
the feed contains fibers, suspended solids, or other debris, pretreatment is necessary. The closely
spaced flow channels will not allow the treatment of viscous liquids. The high pressure drop
through these modules necessitates the use of auxiliary pumping to repressurize the feed between
modules (20). If conditions permit, spiral-wound modules are preferred over tubular units be-
cause this configuration has larger membrane surface area and productivity (20, 27).
The hollow fiber module gives the highest productivity and lowest basic equipment cost
because it has the highest surface area per unit volume of any of the modules (1,19, 26). The basic
equipment cost is low because the individual fibers can withstand the high pressure without
support and are simply encased in a pressure vessel. However, this configuration is the most
susceptible to fouling and concentration polarization, and the external flow type is the hardest to
to clean (9, 20, 21). The units cannot be used to treat viscous liquids, or those containing sus-
pended solids (unless pretreated). The newer internal flow fibers have a better resistance to fouling
and can be backwashed. Hollow-fiber modules are most widely used for demineralization, rather
than waste-water treatment.
Tubular modules are an attempt to solve operational problems at the expense of overall cost
and efficiency. Tubular systems, both internal and external flow, give fluxes and rejections
comparable with spiral-wound and hollow-fiber modules, but the productivity per unit volume at
conventional fluxes is much lower because of the much smaller surface area per unit volume (20,
24). Tubular units are the best choice for viscous liquids, and their operation can be adjusted to
reduce the effects of fouling, plugging, and concentration polarization. Cleaning is easiest with
tubular units and higher pressures in the range of 2.8x106 Pa (400 to 1500 psi) can be used, as
opposed to 2.8x106 to 4.1x"106 Pa (400 to 600 psi) with spiral-wound or hollow-fiber units. The
tubular units need the least, if any, pretreatment of feeds but require substantially larger capital
outlays because they are the most expensive to fabricate and to install.
Thus, the choice of module depends heavily upon the intended applications, productivity
requirements, space limitations, and pretreatment capabilities. The lower cdst and higher producti-
vity of spiral-wound and hollow-fiber units may be offset by the cost of pretreatment to prevent
fouling and plugging of these units. Treatment of a viscous feed with a high suspended solids
concentration precludes any but tubular units. It is also possible that extreme pH conditions could
eliminate any of the available membranes, except the inorganic tubular varieties. The module-type \
choice is an important consideration in the design of a waste-water treatment unit.
98
-------
APPENDIX C
ANALYTICAL PROCEDURES
The following procedures were used to analyze the permeates and concentrates from the pilot
plant operation:
Total Solids: Standard Methods1, section 208 A, pp. 91-92
Total Volatile Solids: Standard Methods, section 208 E, p. 95
Ash: Standard Methods, section 208 E, p. 95
Chemical Oxygen Demand: Standard Methods, section 508, pp. 550-554
Bio-Chemical Oxygen Demand: Determined by Vester J. Thompson Jr., Inc. (Chemical, Materials
and Geothermal Laboratories), 3707 Cottage Hill Road, Mobile, Alabama. The Standard
Methods procedure plus a YSI Model 54 RC dissolved oxygen meter was used.
Color: NCASI Procedure2
Total Organic Carbon: Measured on a Beckman 915 A Total Organic Carbon Analyzer plus a model
865 Infrared Analyzer. The only pretreatment was dilution.
Sodium: Determined on a Perkin-Elmer model 303 Atomic Absorption Spectrophotometer with
flame emission attachment. The pretreatment was to dissolve the weighed sample of liquor
in 2 ml. concentrated hydrochloric acid and demineralized distilled water as needed. The
resulting solution was filtered into a 100-ml. volumetric flask using qualitative filter paper.
Filter paper and container were washed several times with water. The solution was diluted to
the mark with demineralized distilled water. If the resulting solution was too concentrated
for analysis, an aliquot was diluted to proper strength with demineralized water.
Total Chloride: Determined by Galbraith Laboratories, Inc., Knoxville, Tennessee
Total Sulfur: Determined by Galbraith Laboratories, Inc., Knoxville, Tennessee
1 Standard Methods for the Examination of Water and Waste Water. American Public Health
Association, et. a/, Fourteenth Edition, Washington D. C., 1976.
2 An Investigation of Improved Procedures for Measurement of Mill Effluent and Receiving
Water Color, Stream Improvement Technical Bulletin. No. 253. National Council of the
Paper Industry for Air and Stream Improvement, Inc., New York, New York, 1971.
99
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-79-209
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
ADVANCED FILTRATION OF PULP MILL WASTES
5. REPORT DATE
December 1979 issuing date
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
John T. McKinnon
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
International Paper Company
Science and Technology DivisionERRL
P. O. Box 2787
Mobile. Alabama 36601
10. PROGRAM ELEMENT NO.
1BB61Q
11. CONTRACT/GRANT NO.
R 803667-01-1
12. SPONSORING AGENCY NAME AND ADDRESS
Industrial Environmental Research Laboratory
Office of Research and Development
U. S. Environmental Protection Agency
Cincinnati, Ohio 45268
13. TYPE OF REPORT AND PERIOD COVERED
Final 4/28/75-2/27/18
14. SPONSORING AGENCY CODE
EPA/600/12
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Laboratory and pilot plant studies of reverse osmosis (hyperfiltration) and ultrafiltration of pulp mill
wastes were performed by International Paper Company and Oak Ridge National Laboratory (subcontrac-
tor). Decker filtrates were treated with dynamically formed reverse osmosis membranes consisting of
hydrous zirconium (IV) oxide and polyacrylic acid. Bleach plant caustic extraction filtrate (Ej) was treated
with dynamically formed ultrafiltration membranes consisting of hydrous zirconium (IV) oxide and
hydrous silicon (IV) oxide.
The units demonstrated that a properly designed system could effectively treat pulp mill wastewaters
to reduce pollution loads and provide recycleable permeates. Mechanical difficulties prevented the
accumulation of sufficient data for seale-up calculations, but the experience obtained will be useful to
future work involving dynamic membrane systems.
17.
KEY WORDS AND DOCUMENT ANALYSIS
a.
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
circulation*, color, membranes*, osmosis*, pulp
mills*, waste water, water pollution
Hyperfiltration, Mott porous
steel tubes, permeates, reverse
osmosis, Selas porous ceramic
tube bundles, ultrafiltration,
Union Carbide porous carbon
tubes, UCARSEP
13B
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. Or rAuca
112
20. SECURITY CLASS (Thispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (R«». 4-77) PREVIOUS EDITION is OBSOLETE
100
« U.S. GOVERNMENT PRINTING OfTOt: 1MO-657-146/5560
------- |