PB 91-242420
EPA/600/9-91/033
September 1991
MEMBRANES FOR DRINKING WATER TREATMENT
Benjamin W. Lykins, Jr., Chief
Systems and Field Evaluation Branch
Drinking Water Research Division
Risk Reduction Engineering Laboratory
Membranes Workshop
Cincinnati, Ohio
August 6, 1990
RISK REDUCTION ENGINEERING LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 45268
-------
II
NOTICE
This document has been reviewed in accordance with
U.S. Environmental Protection Agency policy and
approved for publication. Mention of trade names
or commercial products does not constitute endorse
ment or recommendation for use.
-------
ABSTRACT
Several Federal regulations have been promulgated and many more are expected
for limiting the concentrations of contaminants in drinking water. As these
regulations are developed, Best Available Technology (BAT) has to be stipulated
for meeting these regulations. Of special interest to EPA at this time is the
control of disinfection by-products while maintaining an adequate disinfection
residual. By 1995, a regulation for disinfection and disinfection by-products is
expected to be in effect that will impact virtually every water treatment plant
in the United States.
Various treatment technologies have proven effective in controlling
halogenated disinfection by-products such as precursor removal and the use of
alternative disinfectants. One of the most promising methods for halogenated by-
product control includes removal of precursors before disinfection. Research
studies in Florida indicate that membranes are effective in removing halogenated
by-product precursors from certain waters. However, can membranes be used as
effectively in other locations for most drinking waters so that they can be
considered BAT for disinfection by-products? There are also other regulatory
concerns where membranes can provide adequate treatment. Membranes can be used
for removing inorganics and radionuclides. Also, with appropriate pilot-scale and
field-scale data, membranes could possibly be considered BAT for meeting the
Surface Water Treatment Rule requirements.
In order to discuss some of these issues, a membrane workshop was held in
Cincinnati, Ohio on August 6, 1990 at EPA's Andrew W. Breidenbach Environmental
Research Center. The major focus of this workshop was to find out what research
has been or is being done with membranes relative to drinking water, future
research anticipated, types of membranes available and expected in the future, and
the possibility of membranes being designated as BAT for future Federal
regulations. To accomplish this objective, several membrane manufacturer
representatives, university and consultant researchers, American Water Works
Association, American Water Works Association Research Foundation, and EPA's Office
of Drinking Water were invited to present appropriate information and data.
The enclosed document is a proceedings that was produced by taping and
transcribing the workshop. Presentations with questions and answers are included.
Some of the presentations contained preliminary data which should be used with
caution unless verified by the presenter.
///'
-------
AGENDA
MEMBRANES FOR DRINKING WATER TREATMENT
Monday, August 6, 1990
Room 130/138
Andrew W. Breidenbach Environmental Research Center
26 W. M. L. King Drive
Cincinnati, OH 45268
8:30-9:00
9:00-10:30
10:30-10:45
10:45-12:00
12:00-1:00
1:00-1:15
1:15-1:30
1:30-2:00
2:00-3:00
3:00-3:15
3:15-4:15
4:15-4:30
Introduction
Treatment Research Summary
(15 min. for each presentation)
Break
Treatment Research Summary (cont.)
Lunch
AWWARF
AWWA
ODW
Open Discussion for Membrane
Manufacturers (what membranes
are available and what to expect
in the future)
Break
Continue discussion
Wrap-up
B. Lykins
G. Amy
B. Cluff
J. Jacangelo
M. Clark
J. Taylor
C. Fronk
M. Weisner
E. Kawczynski
F. Pontius
S. Clark
B. Lykins
|V
-------
LIST OF ATTENDEES FOR MEMBRANE WORKSHOP
Dr. Sary L. Amy, Professor
Department of Civil, Environmental,
and Architectural Engineering
University of Colorado
Boulder, CO 80309
(602) 621-2423
Steve Ary
Research and Development
Systems Technology Div.
Hydranautics
11111 Flintkote Avenue
San Diego, CA 92121
(619) 455-0760
Robert A. Bergman, Project Manager
CH2H Hill
7201 N.W. 11th Place
P. 0. Box 1647
Gainesville, FL 32602
(904) 331-2442
J. Keith Carswell, Sanitary Engineer
Systems and Field Evaluation Branch
Drinking Water Research Division
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
26 W. Martin Luther King Drive
Cincinnati, OH 45268
(513) 569-7389
Dr. Mark Clark
Department of Civil Engineering
Newmark CE Lab-MC 250
University of Illinois
208 N. Romine Street
Urbana, IL 61801
(217) 333-3629
Dr. Robert M. Clark, Director
Drinking Water Research Division
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
26 W. Martin Luther King Drive
Cincinnati, OH 45268
(513) 569-7201
David Elyanow
Ionics Inc.
65 Grove Street
Watertown, MA 02172
(617) 926-2500
Jack Jorgenson
National Water Supply Improvement
Association
P. 0. Box 102
St. Leonard, MD 20685
(301) 855-1173
Stephen W. Clark
Office of Drinking Water
U.S. Environmental Protection Agency
Waterside Mall, Rm. EB55C, WH-550D
401 M St., S.W.
Washington, DC 20460
(202) 382-3028
Dr. Brent Cluff
Geology Building, Rm. 318
University of Arizona
Tucson, AZ 85721
(602) 621-7607
William J. Conlon, P. E.
Vice President
Stone & Webster Water Technology
Services
150 South Pine Island Road
Ft. Lauderdale, FL 33324
(305) 476-1823
Dr. Francois Fiessinger
Vice President, Marketing
Xenon Environmental
845 Harrington Court
Burlington ON L7N 3P3
CANADA
(416) 639-6320
Gerald Foreman
Research & Development
Fluid Systems Division
UOP Fluid Systems
10054 Old Grove Road
San Diego, CA 92131
(619) 695-3840
Carol Ann Fronk
Environmental Engineer
Organics Control Branch
Drinking Water Research Division
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
26 W. Martin Luther King Drive
Cincinnati, OH 45268
(513) 569-7592
Dr. James A. Goodrich
Environmental Scientist
Systems and Field Evaluation Branch
Drinking Water Research Division
Risk Reduction Engineering Laboratory
U.S. Environmental Protection Agency
26 W. Martin Luther King Drive
Cincinnati, OH 45268
(513) 569-7605
-------
Or. Joseph G. Jacangelo
Research Engineer
JM Montgomery Engineering
250 N. Madison Avenue
P. 0. Box 7009
Pasadena, CA 91109
(818) 796-9141
Elizabeth Kawcznski
AWWA Research Foundation
6666 W. Quincy Avenue
Denver, CO 80235
(818) 796-9141
Larry Lein
Desalination Systems, Inc.
1238 Simpson Way
Escondido, CA 92025
(619) 746-8141
Allyn R. (Terry) Marsh, III
Development Leader
Dow Chemical
Separations and Process Systems
1691 North Swede Road
Larkin Laboratory
Midland, MI 48674
(517) 636-6786
David J. Paulson, Manager
Research & Development
Osmonics, Inc.
5951 Clearwater Drive
Minnetonka, MN 55343
(612) 933-2277
Dr. Hermann W. Pohland
Technical Consultant
Permasep Product
E. I. duPont De Nemours Co.
21 Orchard View
Chaddes Ford, PA 19317
(215) 388-2812
Fred Pontius
American Water Works Association
6666 W. Quincy Avenue
Denver, CO 80235
(818) 796-9141
Vernon Snoeyink
University of Illinois
Newmark Civil Engineering Laboratory
208 North Romine Street
Urbana, IL 61801
(217) 333-4700
Alan A. Stevens, Director
Technical Support Division
Office of Drinking Water
U.S. Environmental Protection Agency
26 W. Martin Luther King Drive
Cincinnati, OH 45268
(513) 569-7409
Dr. James S. Taylor
Professor of Engineering
Civil and Environmental Engineering
University of Central Florida
Orlando, FL 32816
(407) 275-2785
Dr. Mark R. Wiesner
Asst. Professor of Env. Engineering
Department of Environmental Science
and Engineering
Rice University, P. 0. Box 1892
Houston, TX 77251
(713) 285-5129
Nicholas Zelver
Eastern Area Manager
Ionics, Inc.
65 Grove Street
Watertown, MA 02172
(617) 926-4304
V
-------
Dr. Gary Amy - University of Colorado
I would like to talk about some of our experiences in looking at the removal
of dissolved organic matter; DBP precursors, so to speak, by membranes with
particular emphasis on nanofiltration. And I would like to highlight work that's
been done at several scales, ranging from bench-scale to pilot-scale to small -
system work. Bench-scale work was done using a groundwater source from the Orange
County, Santa Ana River Groundwater Basin. We used simple test cells to evaluate
the membranes in a bench-scale mode. Next we did some pilot-scale work using
single modules, not meaning to endorse any particular product but this particular
work was done with an NF-70 module. This work was done up in Phoenix, and after
that, I would like to talk briefly about some small systems work that we've done
at a small system in Apache Junction, AZ using multiple modules.
Let's address the bench-scale work first of all. Our bench-scale membrane
test apparatus involves several cells connected in series with the reject from one
serving as the feed to the next. We simply put three square-inch parcels of
membranes in these test cells and run them to steady-state, taking samples of
permeate and reject and evaluating them.
First, I would like to present some of our bench-scale results. Once again,
these were results derived from a groundwater source in Orange County, California.
This is a very colored groundwater, with color as high as 150 color units. Here
we are looking at some results from two different wells, the so called C5 well and
D1 well, looking at untreated waters and nanofiltered waters using a Desal and a
Filmtec membrane. What we see here are reductions in dissolved organic carbon,
UV absorbance which is a surrogate for organic matter, THM formation potential,
TDS and color. We also see some indication of the average molecular weight of
this particular water source. This is a water source characterized by very high
molecular weight humic material, with molecular weights approaching 10,000, so we
expect nanofilters to work very well on this particular water source. This water
source is a groundwater and has very low turbidity, with a turbidity on the order
of about 0.3 turbidity units. You can see that THM formation potentials have been
reduced by roughly about 2/3 using these two particular membranes.
Let's move on to some pilot-scale work that we have been involved with. We've
studied a single NF module in Phoenix, Arizona, looking at two particular water
sources, Colorado River water delivered by the Central Arizona Project aqueduct,
and we've also looked at an effluent that's been recharged though the soil mantle,
and subjected to what's known as soil aquifer treatment as a pretreatment step.
These particular tests were conducted over 30 day durations with an NF 70 module
run at approximately 100 psi. Some selected results from these pilot-scale experi-
ments are shown here. First of all, looking at the Central Arizona Project/
Colorado River water we see a raw, feed, and a product. The difference between
the raw and the feed is we had a slow sand filter as a pretreatment step that did
achieve some modest turbidity reduction. We then proceed from the feed to the
product and you can see reductions in dissolved organic carbon, UV absorbance,
THMFP, and some moderate reduction of bromide; this is a very important issue I
would like to come back to in a moment, the bromide issue. We also looked at
groundwater recharged effluent: there are two wastewater treatment plants that
discharge effluent into a dry river bed in Phoenix. We dug a couple of shallow
wells down-gradient from this particular discharge, pumped the water up after it
had been subjected to some soil mantle treatment, and ran it through a
nanfiltration module. The whole premise here that the soil mantle treatment can
serve as a pretreatment step prior to the actual membrane treatment. Here, it's
kind of interesting that actual effluents coming out of the plants have much lower
1
-------
THM formation potentials than the actual feed into the nanofiltration module. This
is because there's significant ammonia remaining in these particular effluents,
one is partially nitrified and one is virtually non-nitrified, after it goes
through soil mantle. Presumably nitrification happens because the formation
potential of the actual water after it is pumped out of ground and. fed into the
nanofilter is much higher. In any event, there is a 75% reduction in THM formation
potential across the membrane.
Now, one of the things we would like to know is what fraction of the organic
matter is removed in particular experiments, and corresponding with these
particular experiments we subjected the waters to some molecular weight frac-
tionation evaluations to look at the molecular weight of either the Colorado River
water or the actual secondary effluent that is recovered after groundwater
recharge. Here we see before and after nanofiIteration according to various
molecular weight fractions of dissolved organic carbon and THM formation potential.
As we presume, it is the higher molecular weight fractions that are most
effectively removed, and the less than 500 molecular weight fraction is very
slightly if at all removed.
Lastly, we have done some work on small system evaluation. Small system
evaluation has involved a 10 gpm facility at Apache Junction, Arizona. Here, we
are looking at a serial system involving up to 5 membranes in series but have the
ability to manipulate the actual membranes that appear in series. We have the
ability to take samples at various points coming off each module, and we can get
the combined product and the combined reject as well. For this particular system,
I will show you some selected results in terms of the influent to the overall
system and the product, the combined product from the overall system, in terms of
dissolved organic carbon, UV absorbance, and THM formation potential, and here we
see some idea of molecular weight fractionation on what goes in and*what comes out.
If you look at the influent molecular weight of less than 500, a DOC of about 4.9,
and if you look at the product, the overall effluent is very close to 4.7. So we
are really going after the greater than 500 molecular weight organic material.
Last, but not least, I would like to show some results that really look at
what happens over this nanofiltration process train. We have the raw water coming
in, we run it through a slow sand filter, SSF stands for slow sand filtration as
a pretreatment step. Now, we are looking at product waters PI through P5, the
individual modules, CP is the combined product water, and REJ is the reject water.
And one thing I would like to point out on this particular slide is bromide. This
particular system shows only very moderate effects on bromide by this particular
nanofiltration module. I think this is important because if you look at the
trihalomethane formation potential, it is reduced by about 2/3, but if you actually
look at the individual THM species and you express those THM species as THM-Br
verse THM-C1, there is virtually no change in THM-Br across the overall system.
So even though we are affecting THM formation potential, we are virtually not
touching the bromide aspect of this particular situation.
One of my former colleagues at the University of Arizona, Brent Cluff is going
to be talking about this small scale system in more detail during his presentation.
Anyway that's been our experience with removing dissolved organic matter by
membrane processes, particularly nanofiltration.
2
-------
Questions & Answers for Dr. Gary Amy
Steve Ary - Firstly, what is your method, what do you use to measure the removed
organic matter, is there a UV adsorbance measurement there?
Gary Amy - We are looking at three parameters, as with all parameters, dissolved
organic carbon, UV absorbance, and the ratio of UV absorbance to DOC gives us some
idea of the humic content of the organic matter.
Steve Ary - What wavelength?
Gary Amy - 254 nanometers. Then of course, we are looking at THM formation
potential as another indicator of the organic matter in terms of reactivity with
chlorine. So all three of those parameters really address organic matter removal
from different perspectives.
Steve Ary - What is typically, the most problematic molecular weight range, that
THM varies, is it under 500 or what?
Gary Amy - Actually our experience, if you look at molecular weight fractions,
organic matter, is that you find that reactivity actually increases with molecular
weight up to a certain point, and then drops off again. So there is some middle
to high range that is most reactive, some very high range that is not particularly
reactive and the very low molecular weight material is less reactive but still
represents a potential problem.
Steve Ary - So where would your maximum be, molecular weight wise?
Gary Amy - It's really source specific. This Orange County system was basically
material that had a molecular weight of 10,000 or greater.
Bob Clark - I notice on your overhead that you had bromide levels of 80 milligrams
per liter, is that correct?
Gary Amy - It was micrograms per liter.
Joe Jacangelo - When you ran your pilot-scale tests, was that run on a single
module or cascading array?
Gary Amy - The pilot work, the middle work I talked about, was a single module.
Joe Jacangelo - Was there any recirculation?
Gary Amy - No, flow was through a single module with about 75% reject.
Elizabeth Kawczynski - What were the conditions for your THM formation potential?
Gary Amy - They were based on a chlorine to dissolved organic carbon ratio of three
to one, which is the ratio we determined as being necessary to maintain a residual
over a duration of a test, which is a seven day test.
Dave Paulson - Is your work a part of a larger project, and are you looking at the
economics at all?
3
-------
Gary Amy - The economics have been addressed somewhat by my colleague Brent Cluff,
who will talk about that next. I've been mainly involved in chemical
characterization studies.
Dave Elyanow - What were the operating conditions for the nanofiltration modules,
what kind of flux rates, and pressures were you operating at?
Gary Amy - In the last system, the small scale system?
Dave Elyanow - Yes.
Gary Amy - Those were being operated, I believe at about 90 to 120 psi, with flux
rates varying from module to module, with mixtures of individual modules typically
operating about 20 gallons per day per foot-squared.
Ben Lykins - Is there a reason why you selected the membranes that you selected
to do this study? Did you do a screening or any thing like that?
Gary Amy - Some science was involved in looking at product information brochures,
and there was some degree of cooperation involved as well. For people who supplied
us with membranes, we looked at the product information brochures and made some
selections, and people who didn't respond to our request basically didn't get
involved in the first cut so to speak.
4
-------
Dr. C. Brent Cluff - University of Arizona
Nanofiltration was a concept I first heard about while attending a Water
Resources Conference in Kuwait in October 1987. Bruce Watson, associated with the
Stone and Webster Engineering Company in Florida, presented a paper which described
the process (Watson et al, 1987). Bruce Watson called the process membrane
softening. In Arizona we have both a hardness and salinity problem, as well as
a Trihalomethane (THM) problem so we prefer to call membrane softening,
"Nanofiltration" since it is a more inclusive term.
After hearing Bruce Watson's talk in Kuwait, I thought this was something
Tucson needs for the Colorado River water from the Central Arizona Project (CAP)
Water Treatment Plant. After extensive discussions with other nanofilter experts
around the country including Eddie Edwards with the Boyle Engineering Company in
Fort Myers, FL, I was convinced that this was something that could be utilized in
Arizona. The projected costs of $0.50/1000 gallons from Ft. Myers (Edwards, 1988)
for their 20 MGD plant was almost an order of magnitude cheaper than reverse
osmosis systems. The City of Tucson wasn't interested in supporting research since
conventional treatment had already been selected for their CAP Treatment plant.
Fortunately, the University of Arizona was able to obtain research funds from the
John F. Long Foundation Inc. in Phoenix. With these funds two bench studies were
established, one for CAP water and one for recharged municipal wastewater. Both
projects were located in Phoenix.
The results of these bench tests in Phoenix have been documented in two
project reports available from the John F. Long Foundation (Cluff et al, 1989a)
(Cluff et al, 1989b). The organic removal of the nanofiltration projects have been
documented here today by Dr. Gary Amy. Organic removal in these projects has also
previously been reported in the Civil Engineering Environmental Journal (Amy et
al, 1989). A television documentary was completed and shown on Channel 5 in
Phoenix in April 1989 on the recharged wastewater nanofiltration pilot treatment
plant. Copies of this documentary can also be obtained through the John F. Long
Foundation.
In addition to organic removal, the nanofiltered wastewater was also in
compliance with all of the 1986 EPA Drinking Water Standards, both existing and
proposed. The nanofilter was challenged with polio virus and found to remove all
of it.
Using Colorado River water, the nanofilter reduced the hardness (350 mg/L)
by 3/4, the salinity (600 mg/L) by 2/3 and the dissolved organic content (DOC) by
90 percent. This reduction in dissolved organic content also reduced the THMFP
by approximately 80 percent in the CAP water. There were similar results in
nanofiltration of the recharged municipal wastewater.
The results of these bench tests attracted the attention of Consolidated Water
Utilities who sponsored a larger pilot test at the town of Apache Junction, Arizona
using CAP water. The pilot plant is a prototype of a million gallons per day (MGD)
treatment plant Consolidated Water Utilities needs to build for its customers in
the near future.
At the Apache Junction pilot plant, surface water is being used whereas in
Florida, including Ft. Myers, large nanofilter plants are being installed to treat
groundwater. The many nanofilter experts around the country were unanimous in
advising us that we may encounter fouling problems using surface water. It was
5
-------
decided to use slow sand filtration as a pretreatment on CAP water because of
its ability to reduce the organic content in the water through biological action
(Huisman, 1974). It simulates, to some extent, natural recharge. The studies that
were made using CAP water showed that this was a sound approach. Similarly, the
study using recharged municipal wastewater showed that this was also an effective
approach. Wherever practical, recharge basins should be used as the pretreatment
for nanofiltration. In Apache Junction, however, recharge was not practical. This
is due primarily to the excessive depth to groundwater. Slow sand filtration was
selected as a substitute.
Slow sand filtration has been practiced since the nineteenth century. It is
used extensively in Europe where many of the larger municipalities including London
have slow sand filtration systems. New systems are being installed. In the United
States there are very few slow sand systems. In fact, most water treatment experts
in this country know very little about slow sand filtration. In our modern water
treatment textbooks in the United States, the technique is quickly covered with
a few statements such as (1) the method is expensive in that it requires large
amounts of space and (2) it is labor intensive. These statements with regard to
space requirements may have some merit with regard to smaller systems that do not
need any raw water storage, flocculation and sedimentation areas or sediment drying
beds. In large systems that need these areas, there is not much difference in
space requirements. Slow sand filtration serves as the raw water storage, sedimen-
tation and flocculation chambers in addition to the filtration beds. Furthermore,
sediment drying beds are not needed (Huisman, 1974). The filtration rate of a slow
sand filter is generally between 2 to 10 million gallons per day (MGD)/acre.
Cleaning slow sand filters is generally done by removing the top 1/4 to 1/2
inch of sand. This cleaning of slow sand filters, particularly in large systems,
can be quite easily mechanized (Huisman, 1974). In London, an 18 man crew operates
a 400-500 MGD slow sand filter plant. In smaller systems the use of a fabric
matting developed in London appears to reduce the cleaning time. The matting is
dried out, swept off and put back into service. If two sets of mats are used the
slow sand filter can be placed back in service within an hour or so.
Construction costs of slow sand filters can be significantly reduced by using
modern techniques of basins lined with gunite coated plastic rather than poured
concrete basins. Modern techniques of suspended fabric can also be used to cover
the slow sand filters. This covering has been shown to be very effective in
reducing cleaning frequency primarily by eliminating algae growth in the water
above the sand beds (Huisman, 1974).
The slow sand filtration has many advantages as a pretreatment for
nanofiltration. It removes turbidity down to less than that normally obtained in
rapid sand filters. In the cleaning process, only the surface is disturbed so
there is not increased turbidity when the filters are put back into service. If
operated properly and the beds are kept completely saturated then there is a
significant reduction in dissolved organics. In the London slow sand filtration
systems using Thames River water, this reduction is about 25%. Perhaps even more
significant, there is generally a 99.9% removal of micro-organisms that greatly
inhibits biological fouling without the use of disinfectants. Disinfectants such
as chlorine and ozone attack the thin film membranes used in nanofiltration.
6
-------
A patent has been applied for using the combination of slow sand filtration
with reverse osmosis treatment including nanofiltration and ultrafiltration. As
far as can be determined, the University of Arizona was the first to use this
unique combined technology for the treatment of surface water.
The pilot plant facility is shown in Figure 1. It is located next to the CAP
canal east of Mesa. Water is obtained from the canal with a submersible pump.
The water goes through a totalizing meter and then directly into the slow sand
filters. The slow sand filter is constructed out of a plastic-lined redwood box
that is 10x10x8 feet. It has 3 feet of mortar sand on top of 7 inches of gravel
separated with plastic window screen. This type of construction very effectively
keeps the sand out of the gravel so that expensive layering of gravel is avoided.
The effluent is pumped out of a collection well using a submersible pump.
The use of the low pressure nanofilters allows the use of inexpensive submersible
pumps. These pumps are readily available in many different sizes which makes it
possible to optimize the selection.
After the water coming from the slow sand filter is pressurized, it is piped
into a small building where the nanofilter effluent is ozonated. The ozonator is
used in connection with a water dispenser which is a part of the public acceptance
part of the project. Consolidated Water's customers will be allowed to come and
obtain nanofiltered water in their own containers from the pilot plant. The
ozonator was installed not to purify the nanofiltered water, but to purify the
customers containers in case they are contaminated. This part of the project has
not been started yet. The plant has been operating successfully since April, 1989.
The nanofilter train has been expanded in addition to the original three 8-inch
elements. It now includes a 4-inch and a 2.5 inch element. This train is shown
in Figure 2.
The critics to nanofiltration in Arizona were concerned that 20 percent of
the water would be wasted in the reject stream. By adding the 4-inch and 2.5 inch
elements, a "cascading array" was created, whereby, the recovery was increased to
95%. The last two elements were tighter membranes, both SH and SG elements, from
Desal have been used. Presently, SG elements are being used. Both SH and SG
elements have a higher salt rejection than the original Filmtec NF70 or Desal DK
8-inch elements being used. Using tighter elements as the concentration of the
feed water increases is a very important concept in areas where salinity removal
is important.
We have been operating at the 95% level at the Apache Junction Pilot Plant
for the past four months and with no significant change in fouling. Consolidated
Water Utilities have been satisfied enough with the results to start developing
plans for their 1 MGD plant. They have received a preliminary bid on a turnkey
plant to be built by Wastewater Resources Inc. in Scottsdale for approximately
$900,000. This cost includes rapid sand filters, covered slow sand filters and
the nanofilter treatment train including a 1 million gallon covered storage
reservoir for the product water. Although the rated design of the plant in the
winter time is 1 MGD, it will be able to produce twice that in the summer. In a
typical water utility, the demand would peak in the summer when the nanofilters
production due to higher temperature would also peak. However in Apache Junction,
the peak demand is in the winter time due to a high influx of winter visitors.
7
-------
Within the larger 1 MGD plant, it was discovered that the 95% recovery based
on the same type of staging in the pilot plant would have one important
improvement. Instead of a flux rate of 1.66 to 2 GPM that is coming out of the
third 8-inch element in the pilot plant (see Figure 2), there would be 37 GPM
coming out the last element in the 1 MGD plant. Most nanofilter experts we have
talked to, including Larry Lien of Membrane Development Systems (MDS) who is here
today have been impressed that there has been so little fouling with such a low
flux rate in the 8-inch elements that has been used. The elements have only had
to be cleaned every 4 to 5 weeks at the pilot plant. If there is ten times as much
flux going through the elements, the cleaning problems would be much less than they
are in the pilot plant. The nanofilter experts that know of our results attribute
our success to the use of the slow sand filter. It is working very effectively.
The project has been a team effort. Dr. Gary Amy is a part of the team that
has been taking care of the dissolved organics. Another important member of our
team has been Dr. Charles Gerba. He has been doing the micro-organism analysis
for the project. Seeding studies of viruses, Bacteriophage PRD-1, on June 21, 1990
are shown in Figure 3. The seeding level we were trying to achieve was 7 log
pfu/mL. This goal was not quite achieved as shown by the top line of Figure 3.
The difference between the sand filter "out" and nanofilter "in" was that the water
passed through a 5 micron filter. The virus levels, however, are very similar
indicating that the 5 micron filter had little effect. However, the slow sand
filter had a very great effect reducing the virus level from 100,000 pfu/mL down
to 100 pfu/mL. The nanofilter system removed the rest of the viruses.
Not content with the above findings, Dr. Gerba, being the good scientist that he
is, wanted to give the nanofilters a greater challenge. The stilling well after
the slow sand filter was seeded with an even greater loading of 100,000,000 pfu/mL
on July 3, 1990. See Figure 4. These results also showed no significant change
while passing through the 5 micron filter. There was a reduction down to 100,000
pfu/mL from the first nanofilter which was an NF70 and a significant reduction down
to less than 10 from the Desal SH membrane. The final product is a composite from
the 5 nanofilters which showed a reduction down to 10,000. This product was
heavily influenced from the results coming from the first large nanofilter.
Without some leakage in this test there should not be any viruses in the product
water even if it is seeded at a very high level. The filters were checked for
leakage and there was none detected.
The conductivity in the first nanofilter, also measured on July 3, was 150
ppm dropping from a raw water conductivity of 600 ppm. This is about the same
reduction we have gotten since the test started and so there does not appear to
be any leakage. The test will be repeated and reported in the literature.
The pilot plant at Apache Junction attracted the interest of the Pine Water
Association in Arizona. This resulted in the University of Arizona receiving a
contract to install a 40,000 GPD system at Pine, Arizona located at 5000 feet
elevation using surface water. At this site, located at the base of the Mogollon
Rim in Central Arizona, there are cold temperature and snow considerations. Figure
5 shows the design that has been approved for use at this location which will be
constructed this fall.
As shown in Figure 5, a fabric covered slow sand system developed in London
will be used. It is an easy way to clean slow sand filters for small systems.
At Pine, there is going to be a cascading array of 8-inch and 4-inch vessels. In
8
-------
the winter time, all will be operated with a 150 psi pump in order to meet the
demand during cold weather. In the summer time, the lower pressure pump will be
used on half of the array to meet the demand in the warmer weather when the
nanofilter production rate is higher due to warmer temperatures.
An automatic inline electric heater will be used in the winter time to make
sure the temperature of the water never drops below 40 degrees fahrenheit. This
heater will not have to be used very often, but is an essential element in the
treatment train. This same type of design would be appropriate for treating
surface water in most of the colder climates in the United States. If groundwater
is used, then the winter time temperature of the water would not differ greatly
from the summer. The entire University budget for the Pine project is approxi-
mately $26,000, showing that the cost of the system is quite reasonable. There
will be a labor contribution from the utility in addition to supplying the metal
building. At the present time, the utility has no filtration, just a sedimentation
chamber and chlorination. At Pine, there are no salinity or hardness problems,
but the utility is concerned about the taste of the water, dissolved organics and
THM's. These are the reasons they are using nanofiltration.
The research on recharged municipal/nanofiltration and slow sand/nanofiltra-
tion systems have shown that these treatment methods will meet the present 1986
drinking water standards and should also meet future regulations that may be
applied. It appears to be the best available technology for use with marginal
quality water to improve taste, reduce hardness, reduce salinity, reduce pollutants
in addition to reducing the health hazards of disinfectant by-products.
9
-------
11/4" Submersible pump in 8"
plastic pipe
Electrical
Service
Line
CAP
Access
Road
Reject
Evaporation^
Pond
X1
Product
Water Line
Drinking ££
Fountain
Ozonator
-W
Raw Water Line
10 x 10 Slow Sand Filter
5 gal.
Dispenser
6' chain link fence
6x15
Concrete
Pad
Gate
Collection Wells
N Access Door
Nanofilters
Gates
G weied Hoad
CONSOLIDATED
WATER UTILITIES
Plot Plan for Nanoflltration
Demonstration, Apache Junction
Designed by CJt Cluff, WRKC
University of Arizona
FIGURE 1
-------
From CAP/Slow sand filter
12 gpm
8" Nanofilters (NF 70, DK, NF 70)
1st STAGE
29%
RECOVERY
8.46 gpm
4.92 gpm
2nd STAGE
42%
RECOVERY
4" Nanofilter (SG)
\\ 1.39 gpm
7.08 gpm
3rd STAGE
72%
RECOVERY
2.5" Nanofilter (SG)
\\ 0.69 gpm
10.61 gpm
TOTAL
95%
RECOVERY
0.4 gpm
f Reject
4th STAGE
50%
RECOVERY
11.31 gpm
5th STAGE
42%
RECOVERY
11.60 gpm Product
Apache. 5 stage pilot plant at approximate average operating conditions.
Designed by C.B. Cluff. WRRC, University of Arizona.
Schematic
HCHJRK2
-------
Apache Junction (PRD-1)
log (pfu/mL)
7
6
5
4
3
2
1
0
0
4
10
Time (hr)
— Sand Filter - IN ~+~ Sand Filter - OUT Nano-fllter - IN
«- Nano-fllter - OUT Nano-fllter - REJECT
FIGURE 3
June 21, 1990
-------
Reduction of Bacteriophage PRD-1
in CAP Water by Nano-Filtration
Number of viruses (loglO pfu/mL)
10
8
6
4
-a
2
0
80
20 30 40 50 60 70
Time (minute)
90 100 110
0
*- Pre Nano Poat 1st Lg Nano Post 2nd Sm Nano
^~ Final Product —Final Reject
Central Arizona Project
Apache Junction
6 Nanometer Filters
FIGURE 4
-------
-------
Questions & Answers for Dr. C. Brent Cluff
Herman Pohland - Did you try any slime producing bacteria in your assessments to
challenge it in terms of fouling?
Dr. Cluff - No, only two virus have been used. We have not tested with seeded
bacteria. That would be an interesting test to see how effective the slow sand
filter is. I think the slow sand filter does an excellent job on removing slime
bacteria so that we have not had this type of fouling problem.
Dave Paulson - I missed whether you had tested the virus removal rates in the
laboratory, or was that on the large scale elements?
Dr. Cluff - The tests were made using the full pilot system three 8-inch elements,
one 4-inch element and one 2.5-inch element. They were large scale tests.
Dave Paulson - So you were testing for 8 hours with a seeded rate high enough to
challenge an entire element.
Joe Jacangelo - On the virus seeding studies, do you have an idea of the mechanisms
of virus passage. I think this is important because you talk about smaller pore
sizes on a membrane than the size of virus itself.
Dr. Cluff - Yes it surprised us. These results are relatively new, I just got
them last Friday, so I haven't had a chance to really discuss it with Chuck Gerba
to determine what the mechanism might be. We're surprised that the virus got
through.
Mark Wiesner - Is there any chance there was some contamination downstream on
that? I notice that your feed starts at zero, but you already had a concentration
in the product. Was your feed really starting on zero?
Dr. Cluff - It must have been the way samples were taken. I will have to check
on it with the microbiology lab. We have made many studies showing the removal
of natural contaminants. We have also made other seeded studies showing 100%
removal of virus.
Nick Zelver - In the proposed plant in the White Mountains, is there a concern
with Giardia? Do they anticipate the removal of Giardia if there is?
Dr. Cluff - Yes, there is a concern about Giardia. As far as Giardia removal is
concerned, we think we will get rid of it just using the slow sand filter if it
is operated correctly. The City of London has recently had Giardia outbreaks in
the few systems where they are using rapid sand filters. They have never had any
problems with the bulk of their system, which is using slow sand filtration.
Dave Elyanow - You mentioned something about a cleaning frequency of three to four
weeks. What are you doing in terms of cleaning, what kind of chemicals are you
using and how are you gearing up for this at the 40,000 GPD plant in the White
Mountains?
Dr. Cluff - Yes, we clean for a half hour to 45 minutes once every 4 to 5 weeks
at the Pilot Plant in Apache Junction. We would expect that our cleaning frequency
in the White Mountains plant would not be any more than that.
15
-------
Terry Marsh - What were your filtration rates in slow sand and did you measure
that silt density index (SDI) of the effluent?
Dr. Cluff - The slow sand flow rates were approximately 20,000 gallons/day. This
is around the 8 to 9 million gallons per acres per day rate.
Terry Marsh - The second question was were you measuring the silt density index
of the filtrate?
Dr. Cluff - No, we have never done that.
Mark Clark - Where did the reject go?
Dr. Cluff - The reject was evaporated on the site.
Larry Lein - I have followed this project fairly closely. I was astounded at the
recoveries they got. The flow dynamics are terrible for a cross flow filtration,
but the recoveries are quite remarkable in the fact that they didn't follow them
down is even more remarkable. Cleaning once every four weeks is nothing at that
kind of recovery under those poor flow conditions. The only thing you can
attribute it to is the slow sand filter. It is obviously doing a marvelous job.
The kind of slimming and pathogens that you normally see on the membranes just
weren't there.
Jim Taylor - I may have missed the raw water quality or your limitation of salt
and calcium carbonate, how did you control that?
Dr. Cluff - We used FLOCONN 100 antiscalent.
Jim Taylor - What was your hardness?
Dr. Cluff - It was 350 ppm in the Colorada River water.
Jim Taylor - High alkalinity also?
Dr. Cluff - Yes, and a lot of sulfate is also in the Colorado River water. All
of these things are very effectively removed by nanofiltration. I think our
studies have shown that nanofiltration is an excellent treatment for Colorado
River water.
Jim Taylor - Did you run this without the slow sand filter at all?
Dr. Cluff - No, these are tests that need to be done for comparison.
Jim Taylor - The treatment rate through the slow sand filter works out to be about
0.1 gallon per minute per square foot of filter. Is this correct?
Dr. Cluff - It sounds about right, we have always worked with GPD values.
Incidently, concerning the cleaning frequency of the slow sand filter, we went
from September to March without any cleaning. In the summer, we had rain storms
on our canal which brings in fine material from exposed upper banks of the CAP
canal. We have to clean the slow sand filters more often in the summer.
16
-------
Bob Bergman - How many membranes did you use? Were you talking individual elements
per stage? If so, maybe you can relate what the flux was to what the system was
operated at. Did you do any normalized flux decline values in calculations?
Dr. Cluff - We really haven't reduced our readings in terms of gallons per day per
square foot of nanofilter. The membranes were producing a total of 20,000 GPD.
We have been operating for about 14 months. Our pressures in order to get to
20,000 GPD have increased about 5 psi over the 14 months. We started out opera-
ting between 100 to 115 psi. As the membranes fouled, we'd go up to about 115 psi.
Now we are operating between 105 - 120 psi. So gradually over the 14 month period,
we have had to raise the pressure about 5 psi in order to get the same flow. I
think we are probably pretty close to being where we should be in regards to the
life of the membrane.
Larry Lein - I have normalized the data, it is about 20 GFD on the average.
Jim Taylor - Do you have a microfilter, a static microfilter in front of the
nanofilter?
Dr. Cluff - Yes, a 5-micron filter.
Jim Taylor - What frequency do you change that?
Dr. Cluff - We only have two 10-inch cartridges that we change about once a month.
Jim Taylor - This method of pretreatment closely simulates the natural percolation
though soil. The Ft. Myers Florida plant used only conventional pretreatment and
a surface water that percolated through the ground before use. No excessive
membrane fouling was experienced. When this water was used straight from the river
at Olga, Florida tremendous fouling occurred. Consequently slow sand filtration
may be the way to go for surface waters.
17
-------
Joe Jacangelo - JM Montgomery Engineering
I would like to present today some of the data that we have been collecting
on ultrafiltration as opposed to nanofiltration that we have been talking about.
We see ultrafiltration as really becoming an alternative water treatment tech-
nology. One of the reasons is because of the various different regulations that
are coming up, particularly the USEPA Surface Water Treatment Rule. We know that
we have to remove, now, 99.99% viruses and 3-logs Giardia. and provide good turbid-
ity removal. There is also going to be a Groundwater Disinfection Rule, and ultra-
filtration plays a potential role in being able to disinfect water; and finally
the Disinfection By-Product Rule. Even though you don't get very good removal of
organic materials which contribute to disinfection byproducts, using ultrafil-
tration may lessen the amount of chlorine or other disinfectants that are neces-
sary; therefore, you would be able to decrease your total production of disinfec-
tion byproducts.
In ultrafiltration, the way to normally run the modules is by crossflow,
sending the water across the membrane to keep the solids loading off it. Those
materials which are smaller than the pore size of the membrane go through and
those which are not (those larger than the pore size) are retained by the membrane
and are ultimately disposed of as concentrate. Most of the hollow fiber ultrafil-
tration membranes we tested had pore sizes in a range of 0.01 to 0.03 microns.
All the work that we have been doing is performed at pilot-scale. This is
a slide of a typical pilot plant that we would use. You can see here we have a
control panel; everything is run by a program logic controller where we can set
backwash frequency and duration. We also have a number of different valve configu-
rations so we can run the water through the pilot plant in various different ways.
This is a hollow fiber configuration. Here we send the feed water down through
the membrane and the water permeates out the side. Here is just another shot of
an actual on-line system in Concord, California. Again, the feed water tank is
here as is the product water tank. One thing I just want to point out is that we
do have chlorine addition. We add chlorine to the backwash to control biological
and organic fouling; I know that is a concern of many systems. This is another
shot of one of our pilot systems. This particular water is very turbid and ranges
anywhere from 15 to 75 NTU. We are trying to see what is the potential for trea-
ting those kinds of surface waters. All the work that I will be presenting is on
surface water; here you can see the kind of product water that we can filter.
Right now we are conducting three ultrafiltration studies. One of the studies
is in Concord, California where the main objective is particulate and microbial
removal. In terms of chemical addition, we are not adding any chemicals that ap-
pear in the actual product water. This study is done on two waters in California,
one from Sierra snow melt and the other from the San Joaquin - Sacramento Delta.
Another study, funded by AWWARF, is in Boise, Idaho and, again, the objectives are
particulate and microbial removal. Again, we don't pretreat the water with chem-
icals. The only thing that we do is prefilter the water with a 200 micron pre-
filter. This study has been going for about 6 or 7 months now. Finally, the last
study that we are involved in is another AWWARF funded project in Concord,
California and Ottawa, Canada. We are looking at using ultrafiltration, but this
time the focus is more on organics removal and fouling reduction using coagulants
and activated carbon. A lot of work out of the University of Illinois has shown
very promising results on using a coagulant and activated carbon for removal of
organic materials. We want to try to look at the pilot-scale and determine its
potential.
18
-------
When we run our pilot plants we have a raw water feed, pressurize the system
with a pressure pump, and prefilter it with a 200 micron prefilter. The water then
enters into a recirculation loop, is recirculated via another pump, and passes
through the membrane as product water. That which does not pass through the mem-
brane is recirculated back to the head of the recirculation loop. We also have
the ability to bleed the loop so we can control the concentration of the solids
in the recycled water. We also backwash the membranes. Thus, we control the
solids not only by the crossflow but also by backwashing the membranes at various
frequencies and durations. Again, one thing to point out we add some chlorine to
the backwash water and we backwash it approximately every 30 to 60 minutes.
In terms of transmembrane flux, as opposed to nanofiltration, which may
operate at 20 gfd or something on that range, we are looking on the order of 40
to 100 gfd for ultrafiltration. We feel that we really have to get above 50 maybe
60, 70, or 80 to make this process economically feasible, so we are looking at
quite high fluxes. That is the advantage of the ultrafiltration. This slide gives
you a feel for the operating pressures. You can see here the pressures range
between 0 and 15 psi that translates to applied pressures of approximately 10 to
25 psi. You can see here that the pressures increased once we went off tap water
and actually went on to the raw feed water. Initially, we got a little fouling,
but the membrane actually cleaned itself out and settled down around 6 psi
transmembrane pressure. We then raised the pressure to increase our flow rate.
The pressure settled at about 10 psi with a flow rate of 4.8 gallons per minute.
This slide gives you an idea of the turbidity removals we were getting.
Again, this is for hollow fiber membranes. Initially, the turbidity we ran was
on the tap water from the plant. You can see the very low turbidity which was
under one. When we switched to feed water, we found that the turbidities rose up
to 10 and the recycle water that the membranes actually saw was as high as 70 NTU.
For the permeate, we got a consistent turbidity, in this case between .03 and .04
NTU. So you get very good turbidity removal. What we have been doing is some
microbial removal studies, looking at three types of microorganisms: virus, bacte-
ria, and in the next couple of weeks, Giardia. What I will present here is some
of the virus data. We are using bacteriophage. One of the advantages we think
of using membranes for microbial removal is that we do not have to add additional
chemicals for primary disinfection. The primary disinfection step would ultimately
be the membrane filtration. We may be able to decrease the levels of disinfection
by-products because we do not have to add high concentrations of disinfectants to
remove microorganisms. This slide shows some of the results for the virus seeding
studies that we conducted. We used various flux rates; this particular one is a
fairly high flux rate of 111 gfd, and what we did was to seed MS2 virus to the feed
tank and then filter the water and measure the feed, the prefilter, the recycled,
and the permeate waters. We conduct a series of mass balances, so we account for,
in every step, what we put in the feed. We account for it either in the waste or
the permeate so that we know where the microorganisms are at most times. What this
shows is that we were able to remove approximately 7-logs of virus. We found 7
logs after the prefilter. In the recycle water we were able to recover 7.2 as you
would expect a little bit higher because it is recycled as concentrate, but in the
permeate we did not detect any in this particular sampling. At one time, we detec-
ted that one of our filters had a broken fiber. We conducted a bubble test and
were able to isolate that fiber. We seeded it after we found out that one fiber
was broken and noticed an increase in particle counts, and a little increase in
turbidity. We immediately did a bubble test and found that there was a broken
fiber and so we did a seeding study. This slide shows what we found. The feed
19
-------
was about 7.4 logs, prefiltration was 7.6 logs, the recycle was 7.7 logs, but the
permeate now instead of good removals, we found 6.5 logs of virus. It demonstrates
that the integrity of the membranes is extremely important when you are doing
microbial removal studies. As long as your integrity is intact you can get good
virus removal. However, much more work needs to be conducted to demonstrate to
what extent viruses are removed.
This slide shows removal of organic materials. For this study, there are a
series of three different membranes. We looked at three hollow fibers. This is
on Delta water with a TOC range from about 2.8 to 8.9 depending on the time of
year. You can see with these particular membranes, which have molecular weight
cutoffs between 50,000 and 500,000, you get about 20% removal of the TOC.
This slide shows water which has a TOC of 1.7 to 2.8 mg/L, which is substan-
tially less than what we found on the other California water. Again, with the
hollow fibers that we looked at, the molecular weight cutoffs were about 100,000.
You still get very low TOC removal, between 10 and 13 percent. We did look at some
spiral wound membranes. With a polysulfone spiral wound, we were able to get 40%
removal. The molecular weight cutoffs were around 5,000 to 10,000. With the thin
film composite spiral wound, we got around 60% removal.
This slide provides you with an idea of what the SDS and TOX removals were
using various membranes that we looked at. In these particular tests we used
SDSTHMs and SDSTOX. They are similar to the THMFP test except they use lower con-
centrations of chlorine. For the hollow fibers, we obtained very little removal
on either water because of the molecular weight cutoff. Again on this one, very
little or absolutely no removal was observed, depending on the particular test and
membranes. On this hollow fiber polysulfone, we saw no removal, but on the spiral
wound we were able to get removals of about 40 to 60 percent of the SDSTHMs and
SDSTOX.
Just to put organics removal in perspective, we are doing some studies using
nanofiltration in Florida. I am not very familiar with these studies, but the
nanofiltration you can see is much more effective for removing organic materials
in this particular source, which is from Palm Beach County. You can see the feed
water color is reduced, TOC is reduced quite a bit, the THMFP is reduced substanti-
ally as well as BDOC, the biodegradable dissolved organic carbon. So nanofiltra-
tion for organics removal seems to be much more effective than ultrafiltration for
removing THMs and TOX.
One of the critical parameters, we feel, to make the process economically
feasible is recirculation rate. This graph, from our studies of the Boise project,
shows the use of energy per net volume in kilowatt hours per 1000 gallons and this
is the instantaneous flux normalized to 20*C. The range we are looking at is
around 50 to 100 gfd. If you can reduce your recirculation rate, you can reduce
your energy consumption. That is the very key in making the process economically
feasible. Right now, we are undergoing the costing effort; we should have those
costs by October or November. If you reduce recirculation, what happens to keep
the solids off the membrane? You increase your fouling, so its a delicate balance
between trying to lower your circulation rates to a point where you can make it
economically feasible, but also to a point where you can keep the solids off the
membranes.
20
-------
In conclusion, let me say that some of the research needs that we feel in
ultrafiltration are: removal of organics with pretreatment and reduction of
fouling with pretreatment. We'll be looking at these with the project that we
are going to start in November, using PAC and alum. Another research need is the
minimization of energy requirements, and recirculation versus dead-end filtration.
If the latter process can be worked out, that would decrease the cost quite a bit.
Finally, a full-scale plant configuration, single-stage recirculation versus multi-
stage cascading array. For hollow fibers, it is probably feasible to build a plant
under one MGD with recirculation. What about a plant that is greater than one MGD?
Can we afford to push 20 MGD of water around in a circle?
21
-------
Questions & Answers for Joe Jacangelo
Bill Conlon - Can you tell me what the molecular weight cutoff of the spiral wound
ultrafiltration membrane used for this project was?
Joe - I believe it was 6000.
Steve Ary - I think I am confused about the terminology used here. You must have
used at least three types of membranes. One was called an ultrafiltration,
nanofiltration and in another case, I believe it was a microfiltration. What kind
of membranes did you use for this study? Cause I think, well probably I didn't
follow too well, it all came out as ultrafiltration.
Joe - They were all ultrafiltration. There was only one slide, a comparison slide
for nanofiltration that was used in Florida. But the ultrafiltration ones were
all hollow fiber in the range of .01 micron, in terms of the pore size.
Bill Conlon - (Could not understand what he said)
Joe - I don't think there's a set cutoff, it depends on who you talk to as to
where ultrafiltration ends and microfiltration begins. Where does ultrafiltration
start and nanofiltration end. There is an overlap and I don't think the industry
has come to grips with a set cutoff.
Dave Paulson - A definition is just going through a main committee ballot of ASTM
which will define ultrafiltration and I have written it. I added qualifiers, which
ASTM doesn't like but there are necessary comparisons to RO and MF as well, because
you really define it by itself.
Bill Conlon - Whose devices were these? Were they manufactured by Memtec, Rohm
& Haas, Lyonaise des Eau, DSI, or Fluid Systems?
Joe - We did initial screening of five or six different membranes. We looked at
LDE, Romicom, Desal, Fluid Systems, and Bioken.
Bill Conlon - At any point did you consider using tubular devices?
Joe - No we didn't, initially we looked at only spiral wound and hollow fiber.
Bill Conlon - Did you look at using any spiral wound microfiIters?
Joe - No we didn't, again we talked to some manufacturers and we asked them about
using 100,000 molecular weight cutoff spiral wound membranes and the response was
that you would probably get less fouling using a low molecular weight cutoff. So
we went with their recommendation to use a 5,000 to 10,000 molecular weight cutoff
membrane.
Bill Conlon - It seems in your studies though that you found spiral wound to
perform better.
Joe - They performed better in terms of organics removal but not in terms of
operational performance. The problem is trying to filter a surface water with a
spiral wound membrane because there is really no way to keep the solids off the
membranes, so you get fouling very very rapidly; even at recoveries of 10, 15 or
25 percent, and even if the channel spacers are very wide we had difficulty keeping
solids off; you've got to keep the crossflow velocity very high.
22
-------
Bill Conlon - In terms of turbidity then, what kind of removal were you getting
by the 200 micron filters?
Joe - Are you referring to the prefilter?
Bill Conlon - Yes, the prefilters. I realize the viruses probably passed through
the prefilters.
Joe - Very little turbidity removal. The particles, we characterized the
particles, which were less than 120 microns.
Dave Paulson - Yet you showed not detectable or below your detection limits for
the virus. What is your sensitivity and do you know and can you comment on that
sensitivity, in relationship to the probable limits the EPA is going to want to
be looking at for municipal water systems?
Joe - The sensitivity was 7 logs. We saw 7-logs in the feed, so we had a 7-log
sensitivity but what we also did was some 100 milliliter samples so actually in
some of the studies, we had 9-logs of sensitivity. In terms of what EPA is going
to do, the Surface Water Treatment Rule is written for 4-logs removal of virus,
and we were getting 7. How that interplays is yet to be seen. However, it
probably will be necessary to add disinfection at some point, even if it is to just
keep a residual in the distribution system.
Vern Snoeyink - Relative to the use of the membranes for the disinfection purposes,
it would seem that if you are designing a system, you would have to design for a
worst case situation, which would include a broken module or broken tube. Is that
right?
Joe - What you would probably do is provide some type of redundancy in your system.
What we did was to go back and plug that fiber, and we got very good virus removal.
Vern Snoeyink - I am concerned about what would be done in a full-scale plant.
It seems like it would be quite easy for a broken fiber to go undetected for some
period of time.
Joe - In terms of turbidity, if there is just a small pinhole you are probably
right; however, an increase in particle counts would show up right away, so if
you can do some kind of measurement of particles, you can see it immediately.
Bill Conlon - In reference to your differentiation between nanofiltration and
ultrafiltration membranes, most of us in the membrane field feel that a true
ultrafiltration membrane will not reject salt.
Joe - They will not reject salts, yes I agree.
Jim Taylor - Did you look at any direct filtration in your study.
Joe - Look at dead-end filtration? No we didn't. The membranes weren't really
set up for that and it's not clear to date if ultrafiltration can be used with
dead-end filtration just because the pore sizes are so small. The microfiltration
work that you've been doing with Memtec looks very promising in terms of dead-end
filtration.
23
-------
Kim Fox - On the slide where you show a broken fiber, the previous slide you showed
the known viruses from the product water and then with the broken fiber 6.4 logs
or something, but I also noticed that the flux on the first one was 110 where on
this one it was only 55.
Joe - That was only during that particular test. That wasn't a change in flux.
Kim Fox - My question is if the flux had been up like in your earlier one would
that virus count been 7 logs, just to emphasis what Vern was saying the need for
protection, one broken fiber there can really cause a lot of problems. I think
if you had a higher flux, the product concentration of viruses would have been much
higher.
Joe - At a higher flux, it may have been, you are already passing 6-logs, 6 or
6.5 logs out of 7 or 7.5 logs that you seeded, so this is high to begin with.
Kim Fox - It should emphasize even more the effect you would have had if the same
flux was used as the previous one. I think you would have had an even higher
concentration.
Francois Fiessinger - I don't think you can rely on no broken fiber ever, its
impossible if you have fibers. It might be an advantage of spiral wound but I
don't think this is a real problem because you could use a disinfectant that's the
best solution. You will use it anyway, so the few viruses, passing through the
membrane would be killed. The use of ultrafiltration membrane for disinfection,
would be silly anyway. In your experiment, did you refer to reductions over the
treated water already supplied in Boise?
Joe - In terms of virus removal?
Francois Fiessinger - In terms of anything.
Joe - No we just did it on raw surface waters, there was no pretreatment.
Francois Fiessinger - In Boise, didn't you have a conventional treatment plant in
parallel with your membrane unit?
Joe - We haven't looked at that, we are looking at some bank filtration or rather
looking at Ranney Collector water in Boise. But up to this date, we have used only
raw river water.
Francois Fiessinger - Yes because again you can see the advantage of membranes
when you compare it to existing treatment. It doesn't do everything but its much
better than what we have.
Bill Conlon - Have you addressed the economics of using the UF membranes for this
application as compared to the current techonology in use?
Joe - We are doing that right now, we should have everything put together by
October or November.
Mark Wiesner - I'm not sure how silly it would be to use ultrafiltration for
disinfection in terms of the problem of a broken fiber or something. At least
these units are automated to the point where there are so many values and so forth
that if you are going to use particle counts downstream perhaps looking at each
one of the modules, it wouldn't be that difficult to have detection on a given
module and shut the entire module down in case the particle count is high.
24
-------
Joe - I agree. In certain cases, it can be used as an effective disinfection
step.
Francois Fiessinger - In drinking water?
Joe - In drinking water, sure.
Francois Fiessinger - I wouldn't rely on it. I would still use disinfection.
Joe - You will have to, you will never get around that.
Elizabeth Kawczynski - Is there a chlorine residual in the influent coming into
the plant?
Joe - There is 0.1 milligram per liter in one water. We find a 0.1 milligram
per liter chlorine residual in the plant.
Elizabeth Kawczynski - You did a chlorine analysis.
Joe - Yes.
Bill Conlon - Did you determine what effect the prechlorination had on the
membranes life, or the project membrane life?
Joe - No we haven't, we haven't looked at that at all and I'm not sure we will be
able to just because the studies do not run long enough to really estimate what
the membrane life will be. We will have to depend on the manufacturers to give
us an idea of what that is.
Dave Paulson - This is really for all speakers; can you comment on when you are
going to publish these and where they will be published? I assume in the Journal
of the AWWA, some of this at least is funded by that group.
Joe - The initial results are already published in last November's Journal for
the particulate and some of the microbial removal. The cost and economics of it
will be presented at the AWWA Specialty Conference in March. And then we will
put it in a peer review journal probably at the same time.
25
-------
Mark Clark - University of Illinois
I will be talking about some experiments that have been going on at the
University of Illinois for about the last two and one-half years. Most of this
work is funded by the company that Francois just mentioned, Lyonnaise des Eaux,
in Paris, France, as well as, an EPA center on the University of Illinois campus.
We studied, as you notice in the title, the use of ultrafiltration to treat
water from a lake near Champaign, Illinois, Lake Decatur, Illinois. The second
objective was to characterize various ultrafiltration membranes. We really started
from scratch: as a lot of you know, there wasn't a lot of guidance in the litera-
ture about what kind of membranes to use with natural waters; so we just found a
supplier, which in this study was Amicon, and started using these membranes. We
also wanted to look at the molecular weight cutoff of the membranes, and the effect
of that on the removal of turbidity, and natural organic matter. We looked at some
pretreatments, namely coagulation and/or powered activated carbon. We tried to
make some decisions about what kind of membrane material is best in natural water
applications, we also looked at the mechanism of membrane fouling.
In this study we used the basic batch Amicon stirred, pressurized cell, and
we collected permeate and weighed the permeate over time.
The first thing that we did was a literature review to characterize the three
Amicon membranes (the PM, XM, and YM). A lot of you may be aware the PM is a
polysulfone membrane, XM is acrylic copolymer and YM is regenerated cellulose. The
most important thing we found after looking at these membranes, was that the PM
is relatively hydrophophic and the YM is hydrophilic. These are some of initial
results for total organic carbon, UV and turbidity. This is essentially the raw
lake water from Lake Decatur: TOC is 6.14, and UV is 4.9. The Amicon membranes
that we looked at are shown here: the XM100 which is 100,000 molecular weight
cutoff, YM100, PM30 was polysulfone (30,000 molecular weight cutoff) and the YM5.
We found that they all more or less gave you the same removal of TOC. That was
kind of interesting, because we thought if we used a smaller cutoff membrane we
should get better results; but that wasn't the case with this water. UV is shown
here, and of course, turbidity is very well removed by almost any ultrafiltration
membrane.
Thinking about this result (where we didn't get any apparent improvement in
TOC removal with lower and lower molecular cutoffs), we decided to do gel permea-
tion chromatography of the raw water. This figure actually has three curves.
We'll focus on this highest curve, which is for the raw water. We found that there
were two apparent peaks in molecular weight distribution: a very small molecular
weight peak somewhere around a 1000, and a much larger peak, the exclusion peak,
maybe around 100,000. So going back to the previous slide, this explains why the
5000 molecular weight cutoff and 30,000 really do no better than the 100,000: the
small molecular weight fraction goes through all membranes. Now of course, if we
went down to even smaller molecular weight cutoff membranes, I think we would
eventually remove this peak; but most of our membranes were sort of in the higher
range. So that's something important to keep in mind.
Particles are very well removed by membranes, including ultrafiltration
membranes. This is a particle size distribution of the raw water, which shows that
almost all of the particle mass is greater than a micron, so obviously, when you
use ultrafiltration membranes, you are going to wipe out and remove all of this
particulate material.
26
-------
So far, we have been just talking about using the different membranes with
raw water. We also investigated several pretreatment configurations and those are
shown here. Here is the situation which we have been talking about, just putting
the raw water on different membranes and analyzing the permeate. We also looked
at raw water followed by batch coagulation followed by batch ultrafiltration; raw
water with powered activated carbon added, followed by ultrafiltration; and finally
a combination of coagulation and powered activated carbon, with 30 minutes of
flocculation and mixing followed by batch ultrafiltration. These are some of our
results for removal of TOC, depending on the pretreatment step. These are the data
I showed before: only about 40% of the TOC removed with just membranes alone.
If we add a coagulant, mix that for awhile, and apply the whole ball of wax onto
the membranes, we remove slightly more TOC. If we use PAC in a high enough dose
and apply that to the membrane, you can of course remove quite a bit of TOC. I
suggest if you use PAC in conjunction with coagulation, you get about the same
removal of TOC. The other important thing here (which Joe emphasized in some
membranes he was talking about) is if you use two membranes of the same molecular
weight cutoff, you get approximately the same response in terms of water quality
(in terms of removal). It's a different story when we get to flux - but the hydro-
philic YM and hydrophobic XM are pretty much equivalent (in terms of water
quality).
Why does coagulation improve TOC removal a little, and why does PAC improve
it a lot? If you go back to our gel permeation results, remember that our raw
water has a lot of organic material at a low molecular weight and a lot of material
at a higher molecular weight. We did the same gel permeation chromatography after
coagulation and after addition of PAC, and of course, this gives you a nice demons-
tration of why those pretreatments work so well at reducing the TOC. You can see
that coagulation removes a little of the TOC, but adding PAC knocks that small
molecular weight peak way down. So of course, if you preadsorbed the organic
material on the PAC and then you put the PAC on the membrane, you get a nice re-
moval of TOC and turbidity. Obviously, since the PAC has a particle size probably
around 20-30 microns, the membrane will remove the PAC very efficiently also.
We were interested in trihalomethane formation potential reduction, and that
information is shown here as the cross-hatched boxes. Also shown is the TOC. 40%
of TOC is removed just using ultrafiltration; if we add coagulant, we remove a
little more TOC; if we go to PAC plus UF, we remove quite a bit more. Only 5-10%
of THMs are removed by the membrane itself; so ultrafiltration by itself (and Joe
showed some similar results) is not so good. Coagulation starts operating on those
peaks in molecular weight -- starts removing those so that in effect, it is remov-
ing the trihalomethane precursors. Of course, if we go to PAC, we can remove up
to almost 90% of the trihalomethane precursors; combined PAC and UF treatment
systems.
The other side of the coin in membrane processes is the flux and the water
production. This slide shows a couple of important things. Remember I said that
in terms of quality, the type of membrane material was not that significant as long
as the molecular weight cutoff was the same. In terms of quantity, the difference
in membrane material can be quite significant. Here we see the importance of using
the YM Amicon membrane as opposed to the hydrophobic XM: much higher fluxes
result. This shows percent of the initial permeability; initial permeability is
100%, so if you come down to 90%, then you have already lost 10% of your permeabil-
ity. What this shows is that regardless of the pretreatment, you can get almost
27
-------
all of your permeability back after backflushing the YM membrane. The other impor-
tant thing is the flux during the test. We find that using either PAC (and
especially) coagulation pretreatment increases the flux during the filtration
procedure. We get much worse behavior with the hydrophobic membrane: a whole
lot of irreversible fouling is seen even with our best pretreatment situation.
I was going to go into modeling flux here, but I don't think we have time
for that. So I am going to skip through this. We used the classic series resis-
tance model to model different components of fouling.
Obviously, the XM materials fouls a lot more seriously; we also found that
the PM fouls a lot more seriously than the hydrophillic YM. As you know, there
are a lot of schools of thought on fouling. We were interested in trying to
isolate what the mechanism of fouling was with our membrane. As you know, there
are gel concentration models, particulate fouling and adsorption fouling models.
We saw large differences between different membrane materials and we had a hunch
that the irreversible fouling was related to adsorption of organic compounds. So
we thought that one way that we could support that argument would be to just take
the clean membrane pieces, immerse them in the lake water (but not pushing any
fluid through it, thereby not pushing particles into the membrane - just essenti-
ally exposing the membrane to the lake material) and take the membrane out at
different times over a period of 12 hours and test the permeability. If by sitting
in the raw water solution the membrane lost a lot of permeability, then we would
have fairly unambiguous evidence that it was a pure adsorption fouling mechanism,
(in other words the organic materials in the raw water would diffuse toward the
membrane surface and foul the pores). This shows you the results: Just as we
found in the constant pressure experiments, the XM materials really foul very
quickly. We call this the static adsorption test. Just due to the static adsorp-
tion, the XM losses something on the order of 50% of its permeability within an
hour, (so its a very good material for adsorbing natural organic material). Again,
the YM is relatively unaffected during this static adsorption. So we have a pretty
good hunch (and pretty good evidence) for attributing fouling to an adsorptive
mechanism.
Conclusions for the Lake Decatur water (like other people I usually qualify
my results, because things are a lot of times site specific): we can expect in
the absence of pretreatment that most organic matter will pass through typical
ultrafiltration membranes. (I think most people have found that.) However, almost
all particulate matter will be retained. UF without activated carbon pretreatment
is not any better than conventional processes for decreasing trihalomethane forma-
tion potential, (remember we were only removing 10 or 15% of formation potential).
However, if you use enough PAC and give it a long enough residence time, you can
get high THM removal (we are using high doses here - something on the order of 200
milligrams per liter - and if you remember, we were allowing 30 minutes mixing
time). So that is another challenge when we go from these little bench-top experi-
ments to true continuous flow: How do you get enough retention for the PAC and
how do you get a high PAC usage? With or without pretreatment, one of the most
important considerations in the filtration of natural waters is the membrane
material. A lot of other people are quickly finding that (if they didn't already
know) hydrophilic membranes seem to suffer much less adsorptive flux loss than
hydrophobic membranes. Lastly, regardless of the membrane material, pretreatment
can sufficiently improve flux and flux recovery after backwashing.
28
-------
That's all on the ongoing studies. (By the way, this has been published in
the same journal that Joe referred. Another article has been accepted and will
be out this Spring in AWWA.)
Just three final overheads to look at ongoing work. This project, which is
funded by the EPA center on the University of Illinois campus, is "Optimization
of Ultrafiltration for Removal of Turbidity and Disinfection By-Products from
Midwestern Lake Water". We are looking at a number of things here. Most impor-
tantly, we are looking at continuous flow hollow fibers. So we are looking at
longer term effects. (Everything I showed you on the first 16 slides was treatment
that occurred over half an hour.) We are looking again at pretreatment optimiza-
tion using coagulants and powered activated carbon. An important thing in contin-
uous flow operation, is how you do backflushing (and how frequently). We have
found that the length of backflushing is also important. We are also measuring
the THM and disinfection by-product removal.
Really nothing is unusual in the schematic of the continous flow process.
After filtration, we collect permeate in a pressurized permeate collection tank.
Our pretreatment is over here: raw water comes into a constant head tank, through
a prefilter, and up into the rapid mix tank, where we add PAC, coagulant, and
adjust pH.
A second ongoing study concerns the ultrafiltration of Illinois groundwater
using PAC pretreatment. Vern Snoeyink and I are co-PI's on this project which is
financed by Lyonnaise des Eaux, Le Pecq., France. Lyonnaise des Eaux sent us one
of their very own modules, so we had quite a head start here. We did have to rig
up our module. We are concentrating in this study on pretreatment using PAC with
the local groundwater. We incorporate the practical questions of how to optimize
backflushing and PAC dose and that sort of thing. We also want to make use of some
of Vern Snoeyink's experience in modeling PAC adsorption in continuous stirred
reactors. This derives from the observation that if you have a high enough recycle
rate between the ultrafilter and recycle tank, you can approximate this as a
continuous stirred tank reactor, and thereby utilize some of the recently developed
CSTR models for carbon adsorption. These are some of our preliminary results.
(This has been going on since about January.) Here is flux decline over a period
of 80 hours, (with no pretreatment. Actually, there is some pretreatment - we have
a green sand filter to remove iron before this.) So we are getting a fairly
important flux decline over a period of 80 hours. What a student showed me (right
before Vern and I left) is that we are getting back a whole lot of flux, upwards
to 90%, just with simple backflushing; so the system is pretty encouraging. The
membrane utilized here is a cellulosic derivative; in other words, one of those
hydrophilic membranes which performed so good in the first study we talked about.
These are some adsorption modeling results for the CSTR.
Finally, a project that is a little tangential to the water treatment field,
is the mechanism of membrane fouling by microbially produced macro molecules.
Bruce Rittmann and I are co-PI's on this. It is financed by a consortium in
Peoria. We are going to actually measure adsorption isotherms on membrane
material, and characterize those, and compare with some of the standard models.
We are also getting very deeply into mathematical modeling of flux decline. We
have already found some important things there. Finally, we want to study the
adsorbed molecules, in other words, take a sheet of the fouled membranes and use
things like Fourier transform Infared spectroscopy, paper chromatography, and some
other methods to try to characterize which type of organic molecules are most
responsible for the fouling.
29
-------
Questions & Answers for Mark Clark
Bill Conlon - These devices you are speaking of are microfiltration or cross flow
device, rather than ultrafiltration, correct?
Mark - That is considered microfiltration, that was the slide I skipped over.
Yes it was either .45 or .2.
Bill Conlon - I participated in an ultrafiltration study which lasted three months,
in 1984, at Melbourne, Florida using a surface water supply. We found that by
opening the concentrate control valve daily, the resultant velocity across the
membrane cleaned the boundary-layer or concentration polarization off the membrane
surface and renewed the flux.
Mark - We are looking at actually reversing the pressure on the permeate side, in
combination with what you are talking about.
Steve Ary - On the chromatogram you showed, was the material that was being
analyzed a positively or negatively charged moiety or perhaps neutral?
Mark - The organic material. Gel permeation chromotrography has been criticized
that way. What encourages us was the really direct correspondence between the
GPC results and what we were seeing with the membranes. In other words when we
used membranes at those cutoffs we removed the materials or it would pass depending
on the size.
Steve Ary - According to your presentation a very loose UF membrane, that will
remove the upper end, which is on the order of 100,000 Daltons is the best you
can do because the low end is centered around 500 Daltons. This, I believe, you
can only remove by an RO membrane.
Mark - In the absence of pretreatment.
Gary Amy - Your GPC results suggest bimodum distribution of organic matter, other
than you and Sontheimer in Germany I haven't seen a lot of that. That's the kind
where you see more of a normal or log normal distribution of organic matter. That
just further emphasis' your qualification about the source specific contaminant.
Mark - There are few experts on GPC on our campus, and I have been trying to pin
them down on why we might get results like that; I have not gotten any good answers
yet.
Larry Lein - Did you ever use PAC on the project water?
Mark - No we didn't. I guess our strategy was that since ultrafiltration is so
good on removing particulate matter, why not use it to remove the PAC. Otherwise,
a second solid separation processes is required.
Jim Taylor - When you have done coagulation in pretreatment and direct cleaning
through the ultrafilter, when you've clean the ultrafilter, can you describe how
you cleaned it, did you get full restoration?
30
-------
Mark - The cake was very interesting. Of course these are disk membranes, and
you can sometimes hold the membrane like this, and a distilled water bottle and
flush it, and the thing would come off as a sheet, - as a continuous sheet of cake
layer. That is something we have seen with hollow fibers too, when we backflush.
The students have started calling it "spaghetti". The material comes out as a
continuous cake layer; they put in little vials and keep it around the lab to show
visitors. We have found after about a month the "spaghetti" starts to dissolve
and forms a little sludge in the bottom of the vial.
Jim Taylor - Did you ever chemically clean the other membranes?
Mark - We bought chemicals but we never tried them. We got some surfactants and
detergents; we are actually going to look at that more closely in the ongoing
studies. Any recommendations?
Jim Taylor - I have had some trouble restoring flux when alum coagulation was used
for pretreatment after going through both acid and base cleaning. We found
acidification to pH 4 following alum coagulation at Ponta Gorda reduced fouling
in that operation.
Dave Elyanow - Why were you using a green sand filter here?
Mark - I think partly by habit, we just finished a study using groundwater with
a floe blanket clarifier. We want to remove the iron, I guess, prior to that
process.
Vern Snoeyink - It was just that we wanted to control the quality going into the
process. The level of iron fluctuated unless it was removed. We needed a stable
quality going into another experiment, and we used the same water for the PAC/UF
experiments. I don't think that UF would require that otherwise.
Mark - It wasn't something that we thought out, totally clearly; we had been sort
of set up to do that, and we thought well perhaps there would be some scaling or
something, and we will keep using the green sand filter, but that would obviously
impact the overall cost terrifically. So it's something that we need to look at
more carefully and obviously we would want to try to get rid of the green sand
filter if we could.
Dave Elyanow - Obviously, also the iron is, can be a coagulant in itself and it
would be interesting to see the interaction between the iron and organics on this
water, just under UF.
Vern Snoeyink - The iron in the groundwater is the reduced (Fell) state, and we
would need to introduce an oxidation step to remove it by UF.
Dave Elyanow - You won't have oxidation, of course if you are going through green
sand you can get some oxidation, you are oxidizing. The question would be then
why not oxidize it before.
Dave Paulson - Why did you settle on, why did you decide on the hollow fiber as
the membrane element to test next?
31
-------
Mark - I think we ran into a problem a lot of young investigators run into; and
that is finding the companies that have small bench-scale cells that are well
publicized, and can be ordered from a catalog. We found that Amicon was the most
successful there. We are always willing to hear about new bench-top systems.
Bill Conlon - Just a suggestion, you might try a polymer for a coagulant aid
rather than alum, because it is a known fact in the ultrapure water industry, that
alum in city water supplies which comes through their filtration system has been
a problem and has caused plugging of membranes.
Mark - How do they determine that? Do you know? Why do they blame alum?
Bill Conlon - The industry has isolated and proven the fact that it is alum in
many cases causing the plugging problems.
Gerald Foreman - When you get into these situations, you have a real bad fouling
problem. Take the elements apart and just analyze the membranes, alum will be
there.
Bill Conlon - You need to match the polymer to the membrane used. Both the
polymers and some membranes are charged and an undesirable reaction could occur.
It is recommended that one check with the membrane manufacturer when selecting a
polymer.
Jim Taylor - It will be pretty difficult to get the same degree of organic removal,
with a polymer on a high TOC water.
Mark - We had not optimized coagulation, in any sense here. In highly buffered
water we could presumably have lowered the pH with acid and removed more of the
TOC.
Bill Conlon - Is there a reason you prefer to use the bench-scale rather than use
a standard production 4 inch membrane? Is it economics? You will obtain more
accurate, "real-world" data by using a commercial, full-production skid with
standard "off-the-shelf" membranes.
Mark - My feeling was, that with only having limited experience myself, and having
young Civil Engineering Masters students coming in, I didn't want to throw them
into a situation, and say: "here, take this pilot system and go out to Lake Decatur
and give me some flux results". I wanted to do a little more fundamental study
of membrane characteristics and molecular weight cutoffs, examining quality of
water so that first of all, I understood some of the mechanisms and underlying
things better. Now of course, we are going into that; we have a little more
confidence in the technology. Two of the studies I showed you are exactly what
you are suggesting - continuous on-site studies.
Bill Conlon - I think you need to use standard production membranes to gather
"real-world" data and at the same time while using the same water supply test
against the current processes being used. Then convert these test results for
both systems into economic data and determine project feasibility.
Mark - We like to do both, frankly: the fundamental things and the pilot studies.
32
-------
Jim Taylor - University of Central Florida
What I am going to do in this presentation is discuss the productivity and
the water quality associated with either trihalomethane or disinfection by-products
in our membrane work at the University of Central Florida. What I am going to show
you is data that will suggest (1) that there are several membranes that can remove
organic precursors that form trihalomethanes or disinfection by-products, (2)
utilizing conventional membrane systems in a highly organic drinking water is a
very sound thing to do for DBP control, and (3) membranes have been shown to be
consistently productive using highly organic groundwater. Treatment of surface
waters by membranes produces very high quality water but is not very productive.
The data in Figure 1 shows membrane selection criteria from the Village of
Golf, Acme Improvement District, Flagler Beach, Daytona Beach, and Pinellas County
which are groundwater sites, and Olga, Ponta Gorda and Melbourne, Florida which
are surface water sites. The purpose of listing these membranes is simply to show
the molecular weight cutoffs(MWC) for the various types of membranes shown in
Figure 1. All these membranes are spiral wound membranes. Membrane selection
criteria was developed using a single element unit which was operated at whatever
pressure was required to achieve 76% recovery. Sometimes recirculation around the
membrane was used. Using such a high recovery on a single element puts a lot of
stress on the membrane and produces a higher permeate concentration than would
operations at a normal single element recovery of 10 to 15%. This method insures
the selected membrane will meet quality requirements in normal operation. We have
recently changed our selection procedures after development of a modeling capa-
bility that allows operation at a lower recovery and prediction of quality at
higher recoveries. We now use 15% recovery. As shown in Figure 1 and 2, a MWC
of somewhere between 400 and 500 is needed to be able to remove in excess of 90%
of the trihalomethane precursors in the membrane feed water. Softening a water
with a hardness of 250 to 350 mg/L CaC03 to approximately 100 mg/L CaC03 requires
a MWC of closer to 300 as shown in Figure 3.
We are going to talk about membrane productivity and water quality at several
pilot plants. Some of this data has been shown before and some of it has not.
The data in Figure 4 was taken from a groundwater pilot plant at the Acme Improve-
ment District in Florida and it shows trihalomethane formation potential using
Filmtec NF50 during more than 1,000 hours of operation. The average THMFP of this
data was around 35 ug/L and was among the first work that we did showing that
membrane processes control organics or THM precursors. The natural formation
potential of this water was 500 to 700 ug/L. The project flux at Acme is shown in
Figure 5. When we first started working at Acme we initially moved the membrane
pilot plant from the Village of Golf and had not cleaned the membranes prior to
Acme operation. As shown in Figure 5, We ran for some 700 hours, cleaned the
membranes and then sustained a relatively low decline in productivity over the
period of the remaining 500 hours. The T0XFP and the percent TDS, total hardness,
and alkalinity removal is shown in Figures 6 and 7. TOXFP was reduced to approxi-
mately 30 ug/L, but the finished hardness was over 175 mg/L CaCO,. The membranes
used in this pilot plant were Filmtec NF-50s and had a MWC of 500 which was too
loose to achieve the desired softening. Following this study we had demonstrated
membranes could be used as a productive water treatment system producing high
water quality using a highly organic raw water source.
33
-------
The system flux is shown in Figure 8 for the Flagler Beach membrane pilot
plant during some 5,000 hours of operation. We averaged a flux of about 13.3
gsfd, had very systematic operation, and established a productive and stable mem-
brane system. The membranes were Filmtec NF-70s which had a MWC of 300. The
organic reduction was very similar to that at Acme but the total hardness was
reduced to approximately 125 mg/L CaC03. This source was a highly organic ground-
water with a formation potential of around 700-1000 ug/L. The non-purgeable dis-
solved organic carbon(NPDOC), THMFP, and total organic halide formation potential
(T0XFP) in the feed and permeate stream is shown in Figures 9, 10 and 11. The
average conditions of operation were 75% recovery and 140 psi. The average
permeate NPDOC, THMFP, and TOXFP was 1.7 mg/L, 17 ug/L, and 34 ug/L CI,
respectively, in the permeate water. Again a very high water quality was produced
from a very high formation potential source with consistent production using
membranes.
The remaining groundwater membrane systems show DBPFP reduction at Daytona
Beach (DB) and Pinellas County (PC), FL. The flux and mass transfer coefficient
(MTC) is shown in Figures 13 and 14 for DB. The pressure and recovery were system-
atically varied which resulted in a flux variation of 22 to 12 gsfd. The MTC
variation is because of the setting of new membranes intially and membrane fouling
at 4800 hours. This fouling was caused by a new well field that was placed on line
during pilot plant operation. The membranes were cleaned by acid and base solu-
tions and were partially restored. Flux can be expected to decrease with opera-
tion due to deterioration of the membrane. A permanent MTC reduction of approxi-
mately 10% occurred during the DB operation. It is not possible to allocate the
decline between the well field fouling and the natural deterioration. The NPDOC,
THMFP, TOXFP, and DBPFP concentrations in the feed and permeate streams are shown
in Figures 14, 15, 16, and 17 which averaged 0.3 mg/L, 36 ug/L, 43 ug/L and 19
ug/L CI, respectively. The DBPFP distribution in the feed and permeate streams
was similar; THMs accounted for approximately 60%, haloacetic acids accounted for
approximately 30% and the chloral hydrate accounted for approximately 10% of the
permeate water quality. We are operating a second membrane pilot plant at Pinellas
County and monitoring DBPFP reduction. PC is considering building a 100 million
gallon/day membrane plant in the next four years. The system flux, MTC, NPDOC,
THMFP, TOXFP, and DBPFP for the PC membrane pilot plant is shown in Figures 18,
19, 20, 21, and 22. This is a two stage system with (9) 4"x40M elements (Dupont
A15S), whereas the DB system was a three stage system with 21 similar elements,
Filmtec NF-70s. The flux was stabilized for the two operating conditions as shown.
The MTC rate of decline was reduced from approximately 0.00033/day to 0.00011/day
by changing from 80% to 60% recovery. We thought the decreased recovery would
allow a higher velocity across the membranes, reduce concentration polarization
and MTC decline. That has happened so far. These rates of MTC decline would
require a membrane cleaning frequency of 1 or 3 months if the maximum allowable
flux loss was 10% and feed pressure was not increased. Such cleaning frequencies
are acceptable for membrane plant operation. The average NPDOC, THMFP, TOXFP, and
DBPFP in the PC permeate stream were 0.3 mg/L, 21 ug/L, 43 ug/L, and 21 ug/L CI,
respectively. The distribution of DBPs was approximately 60% THMs, 30% haloacetic
acids and 10% chloral hydrate at both sites. The work at DB and PC has demonst-
rated that membrane pilot plants using nanofiltration or softening membranes can
operate consistently producing low DBPFP water from sources with high formation
potentials.
34
-------
What I have shown up to this point has been productivity and quality in
membrane pilot plants that used groundwater sources. I also want to talk about
productivity and quality at surface water plants. We have three membrane pilot
plants that have used surface water; Punta Gorda, Olga, and Melbourne. In each
surface water treatment plant we have a relatively high formation potential source.
You can make the formation potential about as high as you want depending on the
chlorine dosage utilized. The problem with all our surface water sources has been
in terms of productivity, not quality. The flux decline at Olga is shown in Figure
24 and was 0.05/day, more than two orders of magnitude higher than the flux loss
at the groundwater sites. Only sand filtration and microfiltration was used for
pretreatment at Olga. The system flux and MTC are shown in Figures 25 and 26 for
the Ponta Gorda membrane pilot plant operation. Each of the large peaks on these
figures signifies chemical cleaning. The smaller peaks signify a microfilter
change. Twenty chemical cleanings were conducted and the microfilter was changed
weekly although it could have been changed daily. As shown on the figures several
different methods of pretreatment were used. In addition to acid or anti-scalent
addition and microfiltration, there was sand filtration, a disbursant-antiscalent
addition, alum coagulation-settling-sand filtration, and alum coagulation-
settling-sand filtration-acid reduction to pH 4 for A1(0H)3 solubilization. Of
these the last method was the best, but we finally used a two week cleaning fre-
quency, a maximum of 50% recovery, a flux rate of 5 gsfd for our recommended design
criteria for our cost estimate. We don't feel an acceptable method of using mem-
branes on highly organic surface water sources has been developed yet.
Each of the surface water sources has TOC that range from 15-40 mg/L, color
from 50-450 CPU and hardness that gets as low as 50 in the rainy season and as
high as 250 to 300 in the dry season. The low organics and the low color accompany
high hardness and a higher alkalinity because of groundwater intrusion in the dry
season. The feed and permeate stream NPD0C, THMFP, and T0XFP for the Ponta Gorda
pilot plant are shown in Figures 27, 28, and 29, and averaged 0.6 mg/L, 37 ug/L,
and 57 ug/L, respectively, in the permeate stream. That represents an average
organic reduction in excess of 95% and good water quality from a relatively un-
productive pilot plant.
The data I have shown you has demonstrated that several different membranes
can successfully reject DBP precursors from high organic sources, membrane systems
have been shown to produce very high quality water using highly organic surface
and groundwaters, membrane systems are consistently productive in applications
treating highly organic groundwaters, and membrane systems are consistently
unproductive in applications treating highly organic surface waters. Thank you
for your attention.
35
-------
TRIHALOMETHANE AND DISINFECTION BYPRODUCTS (uo/U FOR VARIOUS
SINGLE MEMBRANES OPERATED AT 75% RECOVERY AT SEVERAL LOCATIONS IN FLORIDA
OLGA VOG AID F B P G D B PC MELB
REC =15%
THMFP THMFP THMFP THMFP THMFP THMFP DBPFP THMFP DBPFP THMFP DBPFP
MEMBRANE
Moi. WL Cutoff
FS TFC-4021 76 86 5 13 16 20
50
FS TFCL-4821 18 93
100
FT BW-30 140 21 32 28 76 4 13 6 21 4 11
100
DS D-3 5 42 4 54
150
DSD-5 16 39 15 17
150
DuPTA-15 33 90 33 62 62 108
150
DuPT A-15S 10 26 19 42 35 114
250
FT NF-70 61 88 14 40 15 25 68 120
300
FT NF-50 102 31 39 54 263
400
DuPTA-87 21 82 46 73
400
HN 216 255
DSU90-G1 326 90 326 297 1455 115 359 88 97 193 365
2,000
FT UFP40 707 780
10,000
0S411TS 107 131
10,000
DS U90-G5 500 60S
20,000
OS 411THC 326 97 119
20,000
OS 411 PS 1210 929 370 1363 190 500 68 98 754 1639
20,000
0S411TP 1840 942 376 1084 94 114
40,000
Figure 1
36
-------
96.fi 95.3
Averaoe tew TT-je Calcr
L. „
j-.a csu
29.9
log K.W.
fc^noe tew DOC
20.T bb/L
9
i
*: I
S S
log M.W.
i
-------
so
80 .8
0)
60
>63.6
l- S3
O >
«-> o
JS s
U V
c£
tn
u
e
u.
« a
— >
o
E
a v
c£
O
H
40
20
0
100
80
-< 60
40
20
100
>* 80
•*h e
co 60
ts u
•* » 40
< «
cs
- 20
0
Avg. Raw Chlorides
¦ 93 mg/L
,6.2
_! L
U)
log MW
.79.8
16.7
11.4
Avg. Total Hardness
« 228 mg/L CaC03
log MW
90.8
76.4
Avg. Total Alkalinity
- 131 mg/L as CaCOj
22.2
sa e
f+i
©
u?
I
3 2 CO
v> o
4i e»
a s
1 OH H
* U1 c. c.
-X«
o. o a ¦ o o
"
-------
ISO-i
c
41
O
a.
c
c
o o>
u. s
O)
e
41
E
o
«B
loo-
se.
85-
75-
>»
w
as
>
o
U
41
es «
m *4
3 G5-
O
c.
/¦ • M»
105-
41
k.
3
VI
tft *•
ft? (A
*- CL
100-
"O
41
41
U*
95-
~ II »^.l> I HI
Till
400 600 150 lGuO
Time of Operation (hr)
AID Membrane Pilot Plant Operation Tri ha lonie thane formation
potential versus tine of operation, product recovery rate and
feed pressure.
Figure 4
39
-------
-p»
o
a
M
«/l
"O
M
tx
u
a
•o
M
~J
O.
Ol
21
19
17
15
13
200 400 600 000 1000
Elapsed Time of Operation (Hours)
1200
AID Pilot Plant Operation-Product Recovery
Rate, Feed Pressure and Product Flux versus
TIme of Operation.
-------
15C-i
m
w
e
u
o
a»
c
o
«pv
s
o
u.
VI
41
to
sz
3
e
«
W
i
L-
O
m
*->
©
5(H
91
>
O
u
OJ
V
85-,
75-
U« — 55_
3
•a
a
i-
•• •••• •
200
400
1000
1200
600 800
Tine of Operation (hr)
AID Membrane Pilot Plant Operation - Total organic halide
formation potential versus tine of operation, product recovery
rate and feed pressure.
Figure 6
41
-------
• a*sea on Feed Utter
A Based an lt» Utter
• M Ml I
60
50
40
-¦°\T'NrV"s«AAw~
105-
100 _
95_
200 400 S33 355"
T1«e of Operation
(hr)
AID Msnbrane Pilot Plant Operation - Removal of
total dissolve* solIds, total hardness and alkalinity
versus t1«# of operation, product recovery rata and
feed prtssure.
"SET
Tsfe ifoo
Figure 7
42
-------
16
14
12
10
Q
Li.
cn
O 8
x
D
El 6
4
2
1000 2000 3000 4000 5000
HOURS OF OPERATION
SYSTEM WATER FLUX FOR THE FLAGLER
BEACH MEMBRANE PILOT PLANT
Figure 8 i
l
i
43
-------
AVE. FEED = 10.9
; FEED PERMEATE
AVE. PERMEATE = 1.7
0 1000 2000 3000 4000 5000
HOURS OF OPERATION
NPDOC FEED AND PERMEATE CONCENTRATIONS
AT THE FLAGLER BEACH MEMBRANE PILOT PLANT
Figure 9
44
-------
600
AVE. FEED = 404
! FEED PERMEATE
500
^ 400
300
I- 200
1QQ _ AVE. PERMEATE = 17
G §>- & - -Q „
n *
1000 2000 3000 4000 5000
HOURS OF OPERATION
THMFP FEED AND PERMEATE CONCENTRATIONS
AT THE FLAGLER BEACH MEMBRANE PILOT PLANT
Figure 10
45
-------
1500
AVE. FEED = 1020
i FEED PERMEATE
1200
900
600
300
AVE. PERMEATE = 34
Q L& gy i i ,
0 1000 2000 3000 4000 5000
HOURS OF OPERATION
TOXFP FEED AND PERMEATE CONCENTRATIONS
AT THE FLAGLER BEACH MEMBRANE PILOT PLANT
Figure 11
46
-------
CO
0
X
D
70*
160 lUi
80%
170 pai
80%
150 pst
80%
130 psi
80%
110psi
90%
170 pal
70%
110 psi
70%
170 psi
90%
150 psi
150 psi
HOpsi I'Dpsi
170 psi
2000 4000 6000
HOURS OF OPERATION
8000
SYSTEM WATER FLUX AT DAYTONA BEACH
MEMBRANE PILOT PLANT
Figure 12
47
-------
0.014
0.012
0.010
< 0.008
0.006
0.004
80* : [70* 90*
110pd 1BO psi 170 pi
70*j (70*
110 pi 170 ptf
80%
170 p«i
SO*
150 pw
80*
130 p*J
70* i
0.002 -
70*
170 p*
190*
150 psi
80*
170 psi
0
2000
4000
6000
8000
HOURS OF OPERATION
SYSTEM WATER MASS TRANSFER COEFFICIENT FOR
THE DAYTONA BEACH MEMBRANE PILOT PLANT
Figure 13
48
-------
15
FEED PERMEATE
12
O)
E,
o
o
Q
n.
2
AVE. FEED = 9.3
AVE. PERMEATE = 0.3
€'
2000 4000 6000
HOURS OF OPERATION
8000
NPDOC FEED AND PERMEATE CONCENTRATIONS
AT THE DAYTONA BEACH MEMBRANE PILOT PLANT
Figure 14
49
-------
1000
! FEED PERMEATE
800
- AVE. FEED = 516
^ 600
O)
3
GL
LL
2
x 400
200
- AVE. PERMEATE = 36
2000
4000
6000
8000
0
HOURS OF OPERATION
THMFP FEED AND PERMEATE CONCENTRATIONS
AT THE DAYTONA BEACH MEMBRANE PILOT PLANT
Figure 15
50
-------
1600
1400
1200
^ 1000
O)
3
^ 800
U-
X
O
i- 600
400
200
0
0 2000 4000 6000 8000
HOURS OF OPERATION
TOXFP FEED AND PERMEATE CONCENTRATIONS
AT THE DAYTONA BEACH MEMBRANE PILOT PLANT
Figure 16
51
-------
1000
800
O 600
t/i
CO
s 400
200
!0 HAN £ HALOOeNATED SOLVENTS ThUf ^ CHLOWMATED KETONES
If] CHLORAL MYOBATE (2 MAIOACCTC AOOS ^ FHENOIS
s ~
3
'//I
476
807
1719 2571 3267 3963 4754 5138 6218 7488 8152
HOURS OF OPERATION
80% 70% 80% 70% 80% 70% 80% 70% 90% 70% 80%
170 psi 170 psi 150 psi 150 psi 130 psi 130 psi 130 psi 130 psi 130 psi 110 psi 170 psi
OPERATING CONDfflONS
DISINFECTION BYPRODUCTS FEED AND PERMEATE FOR
THE DAYTONA BEACH MEMBRANE PILOT PLANT
Figure 17
52
-------
24
Q
u_
w
O
U_
80%
140 psi
60%
150 psi
4 -
0
1000 2000 3000 4000 5000 6000 7000
HOURS OF OPERATION
SYSTEM WATER FLUX AT THE PINELLAS COUNTY
MEMBRANE PILOT PLANT
Figure 18
53
-------
0.028
0.024
0.020
% 0.016
0.012
0.008
80%
140 psi
60%
150 psi
0.004
0
0
1000
3000 4000 5000 6000 7000
HOURS OF OPERATION
SYSTEM WATER MASS TRANSFER COEFFICIENT FOR
THE PINELLAS COUNTY MEMBRANE PILOT PLANT
Figure 19
54
-------
5
! FEED PERMEATE
4
AVE. FEED = 3.5
3
2
1
AVE. PERMEATE « 0.2
0
1000 2000 3000 4000 5000 6000 7000
HOURS OF OPERATION
0
NPDOC FEED AND PERMEATE CONCENTRATIONS
AT THE PINELLAS COUNTY MEMBRANE PILOT PLANT
Figure 20
55
-------
300
FEED PERMEATE
250
AVE. FEED = 130
200
-j
O)
3,
Z 150
u_
2
T
100
AVE. PERMEATE = 21
G—-
*©¦
1000 2000 3000 4000 5000 6000 7000
0
HOURS OF OPERATION
THMFP FEED AND PERMEATE CONCENTRATIONS
AT THE PINELLAS COUNTY MEMBRANE PILOT PLANT
Figure 21
56
-------
800
i FEED PERMEATE
AVE. FEED = 490
600
O)
~,
(T 400
~.
x
O
H
200
AVE. PERMEATE = 43
G-
1000 2000 3000 4000 5000 6000 7000
0
HOURS OF OPERATION
TOXFP FEED AND PERMEATE CONCENTRATIONS
AT THE PINELLAS COUNTY MEMBRANE PILOT PLANT
Figure 22
57
-------
300
O
03
CO
250
200
150
100
50
zsz
|G C ~wAoaew'BsscuvBiis Q i»w Q chujwwaud kstonis
|(3 OMUMiw.Hvwu'reg k*us*cet>g aom H cMDmumameHOLS
¦Kftii
mm
//
/ /
¦
6l
\\
137
80%
140 psi
*«ID HJMIA7I HOB MMNMt HD WHMIATt HD PCRMM* MB MMIMt HUB MRMCATI HD WMMfl
1176 14S7 2340 3S04 4618 5423 S546
TIME (HOURS)
80% 80* 80% 80% 80% 60% 60%
140 psi 140 pai 140 psi 140 psi 140 psi 150 psi 150 psi
OPERATING CONDITIONS
DISINFECTION BYPRODUCTS FEED AND PERMEATE
FOR THE PINELLAS COUNTY MEMBRANE PILOT PLANT
Figure 23
58
-------
22
21
20
-a
§? 19
18
17 -
16
15
3 20 40 60 80 100
Elapsed Hours of Operation Since Cleaning (hr)
Permeate Flux at 95 psi feed pressure as a function of
elapsed hours of operation since cfcemical cleaning.
01 ga N -50 Ultrafiltration Study.
120
Figure 24
59
-------
15
12
9
6
3
2nd SAND FILTER
NO ACID
NEW PRESSURE PUMP
ACID ADDED
0
1000 2000 3000 4000 5000 6000 7000
0
HOURS OF OPERATION
SYSTEM WATER FLUX (GSFD) FOR MEMBRANE
PILOT PLANT AT PUNTA GORDA, FLORIDA
Figure 25
60
-------
0.008
0.007
p 0.006
0.005
0.004
0.003
0.002
0.001 - 2nd SAND FILTER
ALUM WATER
NEW PRESSURE PUMP
NO AC10
ACID ADDED ACID ADDED
0
1000 2000 3000 4000 5000 6000 7000
0
HOURS OF OPERATION
SYSTEM WATER MASS TRANSFER COEFFICIENT FOR MEMBRANE
FOR THE MEMBRANE PILOT PLANT AT PUNTA GORDA, FLORIDA
Figure 26
61
-------
25
FEED PERMEATE
AVE. FEED = 13.0
20
15
10
5
AVE. PERMEATE » 0.6
0
0
1000 2000 3000 4000 5000 6000 7000
HOURS OF OPERATION
NPDOC FEED AND PERMEATE CONCENTRATIONS
AT THE PUNTA GORDA MEMBRANE PILOT PLANT
Figure 27
62
-------
1500
FEED PERMEATE
AVE. FEED = 611
900
Q_
U_
5
I
h-
600
300
AVE. PERMEATE = 37
i , r-3-o
0 1000 2000 3000 4000 5000 6000 7000
HOURS OF OPERATION
THMFP FEED AND PERMEATE CONCENTRATIONS
AT THE PUNTA GORDA MEMBRANE PILOT PLANT
Figure 28
63
-------
4200
FEED PERMEATE
AVE. FEED = 2390
3600
3000 "
2400
1800
1200
600
AVE PERMEATE = 57
1000 2000 3000 4000 5000 6000 7000
0
HOURS OF OPERATION
TOXFP FEED AND PERMEATE CONCENTRATIONS
AT THE PUNTA GORDA MEMBRANE PILOT PLANT
Figure 29
-------
Questions & Answers for Jim Taylor
Dave Paulson - Can you comment on what type of membranes, percentage-wise maybe,
the 250 to 300 million gallon a day is made up of, R0, NF.?
Jim - The new capacity? Well you know Bill Conlon could probably answer that a
little better, but I think the majority of those would be the nanofiltration
membranes although there are some brackish water R0 membrane plants being built.
But I might preface that by saying the places that are going to brackish water,
in my opinion, are doing so because of the quality associated with membrane
treatment. I know of a couple of sites that have fresh water sources available
to them that are not as good as the brackish water source in terms of organics and
the quality of the membrane product, whether it is low pressure R0 or nanofiltra-
tion, plays a major role in their decision to go to that process. These membranes
being used in these plants are both cellulose triacetate and thin film composites,
but almost all spiral wound.
Mark Clark - I don't have much experience with R0 but it looks like you were
cleaning every 100 hours or something like that?
Jim - Well the cleaning frequency for the surface water plant at Punta Gorda was
every two weeks. We had never operated at a surface water plant for any length
of time until we got to Punta Gorda and we generally see a 20-25% increase in a
pressure differential or you can see a 10-15% decrease in a mass transfer coef-
ficient before we clean. If we had done that at Punta Gorda we would have cleaned
every other day. We were really trying to identify the effects of recovery, the
disbursant and the cleaning frequency on the rate of flux or MTC loss. We really
didn't develop a system at Punta Gorda that I would ever have recommended. We are
now operating at Melbourne with alum coagulation and we have found that it seems
to work much better than at Ponta Gorda. We are utilizing Dupont A15S membranes
as at Daytona. We are going to do some pretreatment work there with GAC and
microfiltration. We have done that on a smaller scale previously. Both GAC and
microfiltration greatly reduce the flux loss relative to conventional pretreatment.
Microfiltration seems to remove almost all the fouling material at Melbourne. We
are using a Memcor microfilter which uses a cross flow of 10 to 1 and obviously
is expensive to operate. GAC also seems to work effectively. It is interesting
that slow sand filtration and GAC are somewhat biological processes that may
simulate the soil. Another interesting thing along that line is the work that was
done at Ft. Myers, who is building a nanofiltration plant and uses the same water
source that 01 ga does; However, Fort Myers pumps the surface water into canals
which recharge their well field. Using this injected surface water, Fort Myers
had no fouling problems in their membrane study. We had tremendous fouling
problems trying to run raw water and conventional treatment at Olga.
Francois Fiessinger - What is the type of membrane you had in Punta Gorda?
Jim - We had the Filmtec N 70s, a nanofilter.
Francois Fiessinger - The low molecular weight cutoff, 400 explains quality of
your results.
Joe Jacangelo - Was there any bromide in these waters?
Jim - There was not very much bromide, you know there were seven different waters
up there, yes there was some bromide in some of those waters.
65
-------
Joe Jacangelo - The THMs that you found, how were they distributed?
Jim - For THMs where the bromide was present? We didn't measure bromides. The
THM distribution changed in some instances from 80% chloroform to 20% chloroform
with the remaining 80% evenly distributed among the remaining species.
Joe Jacangelo - When you did your THMFP, were they based on chlorine to TOC ratio
or were they all at just one dose?
Jim - Neither. We would establish what chlorine dose we needed to maintain a
chlorine residual at 20°C of at least 0.5 mg/L. You can make THMs almost anything
you want in these waters with a high chlorine dose. We felt that method more
closely approximates THM formation in a distribution system. But that is how we
did that. Typically we would use 5 mg/L for the permeate.
Steve Ary - I think a GP Chromatogram has shown a high molecular weight cutoff.
What would be your tradeoff if you came down to a cutoff of 1000 or 500? Where
would you find your optimum?
Jim - We operated in the neighborhood of 75% recovery, of course in a diffusion
controlled membrane process, the higher the pressure the lower the permeate
concentration and the higher the recovery, the higher the permeate concentration.
A lot of our data indicates that sieving controls precursor removal. I think a
molecular weight cutoff of 400-500 is adequate for THM precursors or DBP precursors
to really see a decrease of 90-95% of these precursors.
Steve Ary - And don't have to drop to 50 ppb or something like that?
Jim - For a molecular weight cutoff, we used a membrane with a molecular weight
cutoff of 2000 and found both TOC and THM precursor removal somewhere in the
neighborhood of 50% and when we decreased the molecular weight cutoff to 500 for
3 different waters, we found 95% to 94% precursor removal so that we concluded
500 MWC was adequate.
Bill Conlon - I just want to make some comments on several things you mentioned.
One important thing that I noted in your presentation was how you had an upset
with the water supply wells. When designing membrane systems, I don't know about
elsewhere, but in Florida, especially with groundwaters, one really needs to
consider a pre-flush system prior to startup, because when certain wells are not
used for a long period of time and then are started up abruptly, the product of
corrosion and biological activity occurring below causes the well, especially in
Florida where there are clay stratas, to add colloidal and particulates to the raw
water. The project hydrogeologist should investigate sealing off or cementing the
clay areas so you don't have colloidal fouling. Some of the past fouling problems
we have had with membranes in Florida may have been attributed to colloidal fouling
due to these clay layers. Secondly, your question about the capacity of installed
brackish water membrane plants versus nanofiltration plants. There was a period
from 1969 to about 1985 when there was a hundred million gallons a day of installed
capacity of brackish water systems including the Key West plant which was about
5 mgd of seawater R0. Since 1985, there has been approximately 200 million gallons
per day of nanofiltration being planned, designed, or under construction. That
information appears in a technical paper in the November 1989 AWWA Journal authored
by Dykes and myself. There are still brackish water systems being designed but
66
-------
I think you are going to see a greater number of nanofiltration systems. The main
problem associated with the design of membrane systems is how are we going to
dispose of the concentrate in the future. Thirdly, I noticed where you had some
fouling on the surface water systems that when you added acid you achieved better
performance, am I right?
Jim - Yes, we also dropped recovery.
Bill Conlon - In Boynton Beach where we performed a 2,000 hour membrane test
demonstration, we had both a thinfilm composite membrane and a cellulose acetate
membrane running side by side. These were full production devices. We really
didn't have a high amount of soluble salts that we really had to control. Acid
was assumed only needed for use with the cellulose acetate membrane so it didn't
hydrolyze. We started out with a thinfilm composite not using any acid whatsoever,
and it fouled almost immediately. We determined that we had to add acid, and once
we added acid, within five minutes the fouling cleared up and the membrane returned
to its normal flux. We think the phenomenon had to do with some sort of charged
organic that the acid was affecting. You may find out that may be the case in some
of your work.
Jim - We did acid feed when we started up in the raw water system to control
calcium carbonate scaling at pH 5. When we solubilized A1(0H)3 colloids we had
to go down to pH 3.5 and 4.
Bill Conlon - You might look closer to determine what the effect really was. Was
the acid really reacting to some charged organics? The other thing that you
mentioned about Ft. Myers not having any fouling. Did they test the membranes
through the wet and dry seasons? Fort Myers main supply is from a surface water
(river) which is used to surcharge a wellfield. They may possibly experience some
organic fouling in the future. I hope not. The Lee County plant adjacent to the
Fort Myers intake use to plug gravity filters regularly with blue-green algae.
Jim - I am talking about the membrane pilot study.
Bill Conlon - I understand.
Jim - That's different if you are talking about the sand filters at the plant.
If you put polyphosphate on to a sand filter it tends to keep calicum carbonate
from cementing or clogging the filter.
Bill Conlon - What I was addressing may be possible at the City of Ft. Myers where
they may have only tested during a dry season or a wet season. More than likely,
it was during a dry season. If the membranes were tested during the full season,
then maybe everything will be okay.
Jim - They ran it for a long time.
Bill Conlon - Then if they did it during wet and dry maybe they are safe. But I
was going to say that we found at the Lee County plant nearby that the conventional
filters would plug-up because of blue-green algae almost every six months back in
the mid-seventies.
67
-------
Jim - The point I was trying to make there, I have had people tell me that anytime
you can use a well system you are a lot better than using a surface water system,
even if you get the intake right next to the river. That just seems like a good
method, researching pretreatment as opposed to the expensive microfilter or whatnot
if you use it.
Bill Conlon - What I am saying is that maybe some of the residual blue-green algae
and organics will pass through to the surcharge well system, even though they are
surcharging these canals and they are pumping out of shallow wells. They still
may have a problem with that.
Dave Elyanow - I just was going to ask about the SDI you are getting in the Punta
Gorda plant.
Jim - We are running at Melbourne right now, but we did do SDIs at Punta Gorda
and we found SDIs between 3 and 4 for the raw water on some occasions even below
3 for the raw water, we found them lower, much lower for the coagulated water.
I have a student whose about finished his thesis who works for Bill Berger that's
done a very good work on pretreatment indices and his work shows clearly for the
waters he's tested that the MFI index is quite superior to the SDI index for
correlating fouling results. The SDIs we found at Melbourne were as high as they
could be for 15 minute test, 6.7 for the raw water. The alum coagulation ones were
relatively low. The work we have done at Melbourne though shows that we are having
trouble getting the membranes to clear using alum coagulation, like I mentioned
they are fouling out quickly. Each time we clean them they progressively foul out
a little quicker. Now we have only been on-line for probably about a month.
Dave Elyanow - Even at an index of 3 to 5?
Jim - At Melbourne the silt density index is 2. It is low. We use alum coagulated
water so I really don't like the SDI index as a means of predicting membrane
fouling. I am hopeful the MFI index will be more useful.
68
-------
Carol Fronk - U.S. Environmental Protection Agency
The purpose of my discussion is to look at membranes for removing
specific organics from drinking water. Sources of water in the United States
consist of surface waters and groundwaters. The types of mixes of contaminants
will be different in different waters. Surface waters may have many organics with
low concentrations and groundwaters will have fewer organics with higher
concentrations. We presently are looking at four different types of compounds:
volatile organic contaminants, synthetic organic contaminants, disinfection by-
products, which have been discussed this morning, as well as inorganics. Examples
of VOC's, or volatile organic contaminants, are solvents like trichlorethylene,
tetrachlorethylene, vinyl chloride and benzene. Examples of SOC's, or synthetic
organic contaminants are pesticides like lindane, carbofuran, alachlor, and
atrazine but also include compounds like PCB's. Inorganics that are commonly
found in water are arsenic, barium, and nitrates. The sources of contamination
are varied: pump lubricants, industrial discharges, dry cleaners, degreasing
agents, as well as dumps. There are two different types of entry into water
systems: 1) point discharges may originate from factories and flow into groundwater
or surface water, and 2) non-point discharges may come from applications of
pesticides on fields that then run off into streams, lakes, and rivers. Specific
examples of the above might consist of 1) pesticide point discharge from a
pesticide producing plant that may directly enter a surface water, or 2) the
application of pesticides onto agricultural fields - when it rains in the spring
the pesticides may be carried into the surface water. In many cases the pesticides
will pass through conventional waste treatment and wind up in the tap water.
The goals of the EPA Drinking Water Research Division, of which I am a part,
are to evaluate treatment technologies for control of organics and provide data
to ODW in support of the regulatory process. We also look at various technologies
for full-scale application. We accomplish this, in two ways, we look at
technologies both in-house, in our laboratory, and extramurally, in field studies.
Membrane technologies have shown promise. Because RO membranes vary in
configuration and polymeric chemical structure, our approach was to look at a
variety of membranes types: cellulose acetate, hoi low-fiber polyamide, and several
different thin film composites. We also looked at a couple of different
configurations (spiral wound and hollow-fiber). Different chemical names for
various membranes were supplied by the manufacturers. Our approach was to spike
a tank with VOC's or pesticides, then using low pressure pumps, pass this spiked
influent water through the membranes and collect permeate, concentrate, and
influent samples.
At present we are running four different types of RO membranes, side-by-side,
in parallel: a Culligan unit which houses a Filmtec FT-30 membrane, a DSI membrane,
a UOP membrane and a PEC membrane, (a Japanese membrane from Toray Inc.). Our
preliminary research shows that for compounds such as cis-l,2-dichloroethylene,
a volatile organic, that the cellulose acetate, and a hollow fiber-polyamide
membrane did not produce very good removals. Nevertheless at the time, there were
indications that some of the thin-film composites were capable of greater than 70%
removal of certain VOC's. This figure shows a comparison of the average percent
removal of four VOC's by several membrane types. As can be seen, thin-film
composite C has the superior average percent removal. The pesticides on the other
hand, such as aldicarb sulfone and aldicarb sulfoxide as well as carbofuran were
all well removed by a variety of different membranes. Longer term tests showed
69
-------
that cellulose acetate membranes gave rather poor removal of VOC's, the hollow
fiber-polyamide membrane a little better, but by far the thin-film composites
showed superior removal. Additionally, some thin-film composites showed better
removal than other thin-film composites for certain classes of compounds. Moving
onto pesticides again we see the cellulose acetate membrane not removing as well
as the hollow-fiber polyamide, but that the thin-film composites performed very
well with over 90% removal for most pesticides. Data collected from other
researchers, also indicate that the UOP and PEC 1000, both thin-film composites,
have a molecular weight cutoff of about 75, and exhibit good removal (over 90%)
of aldehydes, ketones, acids, and alcohols.
After the success of the inhouse research, it was decided to take the mem-
branes into the field and see how they would work. The objectives of this research
were to see how well the membranes could remove organics and to look at some of
the operational and maintenance problems that might exist. In order to do this,
we went to a Northwestern Ohio site where pesticides are applied to corn and soy-
bean crops. During spring run off, these pesticides run into the surface water,
into the rivers and ultimately into the lakes. The compounds of interest in this
case were nitrates and pesticides. This was a joint project in which EPA teamed
up with Heidelberg College and several water treatment plants. Specifically, I
will be addressing the work done at a small town, Tiffin, Ohio. The Sandusky River
runs through Tiffin and then into Lake Erie. The pesticides that are applied to
the watershed runoff into the Sandusky River, and pass through their conventional
water treatment unaffected. Ultimately they find their way into the tap water.
The pesticides that we looked at this site were alachlor, atrazine, and carbofuran
because these are the pesticides that are primarily used in this particular
agricultural area. The pesticides applied will vary throughout the country and
are dependant on the crops grown. Pesticide concentrations had been running
between 10 to 30 micrograms per liter, prior to this study for simazine, atrazine,
alachlor, metolachlor, linuron, and cyanazine. We took three reverse osmosis units
to this site: one unit housing a cellulose acetate membrane, another a hollow-
fiber polyamide membrane, and one housing a thin-film composite (an FT-30). The
study was conducted in 1986 for approximately a month during Spring runoff. This
shows excellent removal of the project pesticides over that length of time by the
FT-30 membrane. Cellulose acetate membranes removed approximately 20% to 90% for
alachlor, atrazine, and metolachlor. The hollow-fiber polyamide membrane showed
very good results, after it had reached steady state, but then there was a decline
in removal toward the end of the study.
To further verify-the field work, we decided to do additional inhouse testing
on these pesticides. The influent was kept at approximately 100 micrograms per
liter, and mass balances were verified for all tests. Mass balances were good,
within +/" 20%. Again for the FT-30, alachlor was well removed, with an influent
of approximately 100 ug/L and a permeate of about 0 ug/L. Linuron was similar with
100% removed, good mass balance verification and an excellent quality of permeate
coming out. Carbofuran removal by a hollow-fiber polyamide ran about 70%, but
Simazine was not as well removed by this membrane. The next two figures show a
kind of a comparative compilation of both the Tiffin field work and the in-house
pilot scale work. As can be seen for both field and in-house work the thin film
composites were superior in removing the pesticides tested.
In a similar study conducted at Suffolk County, New York, R0 was compared to
GAC on well water (Lykins and Baier). Several units were tested on a different
set of pesticides. This area primarily grows potatoes, and therefore the pesti-
cides used and their concentrations (0-60 ug/L) in the well water are different.
70
-------
Using a DuPont hollow-fiber polyamide membrane, over approximately a years time,
aldicarb sulfone, with an influent concentration of about 14 to 30 ug/L was removed
by R0 to give a permeate below 1 ug/L. Aldicarb sulfoxide similarly was well
removed. The compound 1,2-dichloropropane, a volatile organic contaminant, with
an influent of about 20 ug/L was not as easily removed, but carbofuran at 10 ug/L
was removed to below 1 ug/L* A compilation of the data shows that over the one
year period of time, steady-state removal was achieved, and that carbofuran,
aldicarb sulfone, aldicarb sulfoxide, were removed between 80 and 100%, and 60
to 70% removal was achieved for 1,2 Dichloropropane.
Another area of concern is inorganics in water. This is exhibited by the
fact that compounds such as nitrates are found in waters and can cause blue-baby
deaths. Inhouse data from Sorg's group shows that point-of-use removal is quite
good (over 90%) for inorganic contaminants.
The disadvantages of reverse osmosis are: units may have high cost, require
pretreatment, and exhibit possible clogging and fouling. On the positive side it
is effective on a wide variety of contaminants, and the sizes and applications vary
from large units to point-of-use.
In conclusion, some membranes are better than others for removing VOC's and
pesticides, and the removals vary according to polymeric blend and type. The
future (directed towards the manufacturers), should be in the development of new
membranes that will improve the removal of lower molecular weight compounds. In
addition, reverse osmosis shows promise for dual problems such as inorganic and
organic contamination, for example in rural areas where both nitrate and pesticide
problems coexist. More tests need to be conducted on pesticides (over a year or
more time) to look at how well steady-state is achieved and to verify mass
balances. You must ensure that adsorption of these pesticides on to the membranes
will not result in a big slug coming off later. Membranes are very flexible in
that a proper size can be used to match any application, but as was said
previously there's a problem with the disposal of the concentrate or the reject
water. As you've seen, Phase I was to look at a variety of membranes. We picked
the best ones out and then did rigorous long-term side-by-side testing (which is
still being conducted) for well over a year. The next phase, which is to begin
in the fall or winter, will be long-term testing on a multiple-pass unit using the
best membranes from Phase I. The best choice at the present appears to be a UOP
or the FT-30. Additionally, to take care of the concentrate problem we will be
looking at other membrane processes such as pervaporation. This is a process where
you are pulling a vacuum instead of pushing the water through the membrane. If
just the organic contaminants could be pulled out of the contaminated solution,
you would have a very concentrated, small volume of waste and a separate large
volume of clean water. This process might then provide a closed system approach
to treatment.
71
-------
CO
w
<
04
PQ
PJ
04
O
lu
05 ft
g < o
W t?
04 O
0
2
2
cs
Q
72
-------
In General
Surface Water: Many Organics. Low Concentrations
I
I
Groundwater: Few Organ ics,
High concentrations
i
j
i
« i
-------
RESEARCH PRIORITIES - ORGANIC COMPOUNDS
REVISED
DRINKING WATER
PRIORITY REGULATIONS
1. VOLATILE PHASE 1
ORGANIC
COMPOUNDS
^1
2. SYNTHETIC PHASE 2
ORGANIC
COMPOUNDS
3. DISINFECTION PHASE 3
BY-PRODUCTS
-------
o
2
<
0
&
o
g
75
-------
VOLATILE ORGANIC COMPOUNDS
TRICHLOROETHYLENE
TETRACHLOROETHYLENE
1.1-DICHLOROETHYLENE
VINYL CHLORIDE
1X1-TRICHLOROETHANE
1.2-DICHLOROETHANE
CARBON TETRACHLORIDE
BENZENE
P-DICHLOROBENZENE
-------
SYNTHETIC ORGANIC COMPOUNDS
LINDANE CIS-1.2-DICHLOROETHYLENE
2,4-D TRANS-1,2-DICHLOROETHYLENE
SILVEX EDB
ACRYLAMIDE DBCP
METHOXYCHLOR TOLUENE
TOXAPHENE XYLENE(S)
PENTACHLOROPHENOL STYRENE
PCB(S) O-DICHLOROBENZENE
ALDICARB ETHYL BENZENE
CARBOFURAN CHLOROBENZENE
ALACHLOR 1,2-DICHLOROPROPANE
ATRAZINE
HEPTACHLOR
CHLORDANE
EPICHLOROHYDRIN
HEPTACHLOR EPOXIDE
-------
PRINCIPAL FORM OF INORGANIC CONTAMINANTS
CONTAMINATION VALENCE
PRINCIPAL FORM
IN WATER
ARSENIC
BARIUM
FLUORIDE
NITRATE
RADIUM
SELENIUM
+3 (ARSENITE)
+5 (ARSENATE)
+2
-1
+5
+2
+4 (SELENITE)
+6 (SELENATE)
ANION H AsO
ANION H 2 As04
H As042
CATION Ba 2
ANION F 4
ANION N031
CATION Ra 2
ANION H SeOa
Se Oa2
ANION Se 042
-1
-------
Sources of Ground Water Contamination
Pump Lubricants - Minor
Industrial Dischargees
Individual Households
Ground Water Recharge
-------
LU
o
I
LU
(3
CC
<
O
CO
Q
O
Q_
CC
LU
CO
CC
_ LU
o
>-
CC
O
£ w
^cr
LL
80
-------
LU
o
0
a.
1
O
O
2
u_ HI
LL
Q Q
e 8
LU
CC _
Q. CC
CO
CO
LU
O
81
-------
EXAMPLES OF PESTICIDE
CONTAMINATION
DIRECTLY FROM
PESTICIDE PLANT
INDIRECTLY FROM CROPS
~
INTO RIVERS
THROUGH TREATMENT PLANTS
f
INTO TAP WATER
-------
<
Ph
w
0
2
2
04
Q
w
!Z
O §
^ 2
<| CO
w £
co rl
W 3
83
-------
IN-HOUSE RESEARCH
OBJECTIVES: EVALUATE TREATMENT PROCESSES FOR
ORGANIC CONTROL
PROVIDE DATA TO ODW IN SUPPORT OF
REGULATORY PROCESS
IDENTIFY PROCESSES FOR FULL-SCALE
COST-EFFECTIVENESS EVALUATIONS
-------
LU
CO
0 Z
1 <
Q
UJ
i I
O
^ QC
^ <
QC ~
CO
LU
< CC
QC
X
LU
85
-------
5 O
es O
ni ®
CU Q
w
E-h
86
-------
CO
LU
z
<
DC
00
2
LU
2
CO
CO
o
CO
o
LU
CO
oc
LU
>
LU
OC
LU
_ cc
9 3
Z h
< o
£ £
2 P
I- CO
^ -J
gc <
3 o
o =
7 W
O g
o w
z a
- oc
>- LU
S 5
* 23
o
Q.
87
-------
MEMBRANE
TYPE
CONFIGU-
RATION
CHEMICAL
NAME
CELLULOSE
CELLULOSE
ACETATE
SPIRAL
ACETATE
WOUND
(DIACETATE)
&
TRIACETATE
POLYAMIDE
HOLLOW
FIBER
AROMATIC
DIAMIDE WITH
AROMATIC ACID
CHLORIDES
-------
MEMBRANE
TYPE
CONFIGU-
RATION
CHEMICAL
NAME
POLYAMIDE IS
THIN FILM
SPIRAL
CROSS LNIKED
COMPOSITE A
WOUND
& CONTAINS
MEMBRANE
COMPOSITE
CARBOXYLATE
GROUPS
POLYETHYLENIMINE
THIN FILM
SPIRAL
CROSS LINKED WITH
m-TOLYLENE
COMPOSITE B
WOUND
2,4-DIISOCYANATE
MEMBRANE
COMPOSITE
COATED ON A
POLYSULFONE
SUPPORT
-------
MEMBRANE
CONFIGU-
CHEMICAL
TYPE
RATION
NAME
MODIFIED
THIN FILM
COMPOSITE C
MEMBRANE
SPIRAL
WOUND
COMPOSITE
POLYALKENE
ON A POLYSULFONE
BASE ON A BACKING
OF NONWOVEN
POLYESTER
THIN FILM
SPIRAL
COMPOSITE D
WOUND
PROPRIETARY
MEMBRANE
COMPOSITE
-------
SCHEMATIC OF REVERSE OSMOSOS SYSTEM
Headspace
Free
Tank
Pump
200 psig
Pressure Vessel
Membrane
Pressure
Regulating
Valve
Permeate (Product Water)
Concentrate (Reject Water)
-------
REVERSE OSMOSIS TREATMENT
percent removal
water membrane
soc/voc
ug/L
di9t
ground
CA
NA
TFC
c-dichloroethylene
39-67
X
0
20
12-32
c-dichloroethylene
48
X
11
EDB
11-68
X
37-84
1,2-dich loropropane
24
X
4
75
38-88
chlorobenzene
17-56
X
50-87
aldicarb sulfone
47
X
94
96
94-99
aldicarb sulfoxide
39
X
99
99
95-99
carbofuran
14
X
86
99
99
alachlor
100
X
>90
RECOVERY
6-8 18-50 5-16
-------
Membrane Type
Thin FUm
ComposiM
Potyamlde
Cellulose Acetate
20 40 60 80
Average Percent Removal
Comparison of Average Percent Removals*
of Contaminants by Various Membranes
100
Average Percent Removal of Chloroform, 1,1 J-TricNoroothene
cte-1,2-Dlchloroeltiylene, and Trichloroaftfiytene
-------
PRELIMINARY DATA
PERCENT REMOVAL
COMPOUNDS
MOLECULAR
WEIGHT
CELLULOSE
ACETATE
POLYAMIDE
THIN FILM COMPOSITES
FT-30fO(5)
100(9)
BROMOOtCHLOROMETHANE
164
0(1 1/4)
44 (1 1/4)
70(4)
80(4)
40(4)
80 «0<4)
DtBROMOCHLOROMETHANE
208
0(1 1/4)
32 (1 1/4)
78(1 1/4)
—
—
95(1 2/3)
BROMOFORM
252
0(1 1/4)
38 (1 1/4)
81 (1 1/4)
—
—
—
AROMAT1CS
BENZENE
78
0(<1)
18 (<1)
80(9)
90(9)
80(3)—50(5)
98(6)—*0(3)
TOLUENE
92
— •
—
80(4)
90(4)
55(4)
90 60(4)
ETHYLBENZENE
108
10(2)
30(2)
80-100(4)
95(9)
95(5^—60(3)
100(6)—*70(3)
O-XYLENE
108
0(<1)
—
96(4)
95(4)
75(4)
100—00(4)
PXYLENE
106
0(<1)
—
60(4)
90(4)
50(4)
70(4)
CHLOROBENZENE
112
10(2)
10(2)
60-70(4)
80(4)
40(4)
100 (1 2/3)
80—60(4)
O-OtCHLOROBENZENE
147
10(2)
10(2)
60(4)
70(4)
40(4)
80—50(4)
P-DtCHLOROBENZENE
147
0(1 1/4)
—
—
—
—
—
M-MCHLOROBENZENE
147
—
—
50-60 (4)
70(9)
70(3)M0(5)
90(5)—50(3)
BROMOBENZENE
157
10(2)
10(2)
—
—
—
100 (1 2/3)*
-------
PRELIMINARY DATA
PERCENT REMOVAL
COMPOUNDS
MW
CELLULOSE
ACETATE
POLYAMIOE
FT-30
A
THIN FILM COMPOSITES
UOP DSI
B C
PEC
D
PESTICIDES & NITRATES
ETHYLENE DCROMIDE
188
—
__
37 (< 1)
49 (<1)
84 (<1)
—
SIMAZINE
202
20
30(2)
100^ 70
60-80 (2)
100
....
100(1)
—
ALDICARB SULFOXIDE
206
97(1-6)
90(12)
97(1-6)
95(1-6)
—
—
METRMBUZEN
214
20-80
100-- 70
100
—-
--
—
ATRAZINE
216
30
40-60 (<1)
100-«- 70
50-80(1)
100
—
100*60(1)
—
ALDICARB SULFONE
222
84(1-6)
®5 (12)
98(1-6)
94(1-6)
—
....
CARBOFURAN
CYANIZINE
222
241
86(1-6)
60-40(2)
SO
80-90(12)
80(2)
100«- 70
93(1-6)
100
93(1-6)
UNURON
ALACHLOR
249
270
0
0 (2/3)
70
100— 70
60(1)
100-- 70
100
100
100 (1/4)
100(1)
METOLACHLOfl
NITRATES
284
80
74
100- 70
90
100
100(1/4)
95
95
....
ALKENES
CIS-1,2-DICHLOROETHYLENE
97
0(2)
10(2)
20(9)
20(9)
20(9)
80(1 2/3)
6O«0O(9)
TRANS-1,2-DICHLOROETHYLENE
97
0 (1 1/4)
—
0(1 1/4)
—
—
—
TRICHLOROETHYLENE
132
0(<1)
32(4)
30-60 (9)
30-70
60-40
80 -60
TETRACHLOROETHYLENE
166
0 (1 1/4)
20 (1 1/4)
80(9)
90(9)
BO-65
90«60
KEY
BLACK - TIFFIN OHIO RED - SUFFOLK CO. NY. - TESTS NOT CONDUCTED
BLUE - IN-HOUSE (#) - NUMBER OF MONTHS RUNNING
-------
100
80
o
2 go
ill
cc
40
20
% REMOVAL VS.
MOLECULAR WEIGHT OF
ALDEHYDES, KETONES, ACIDS AND ALCOHOLS
\ I
-
[I
'
1
1,1.1,!,
50 100 150
MOLECULAR WEIGHT
200
250
UOP PEC-1000
¦¦ o A
-------
EXTRAMURAL RESEARCH
OBJECTIVES:
EVALUATE TREATMENT PROCESSES FOR
* ORGANIC CONTROL
* OPERATION AND MAINTENANCE
* SECONDARY EFFECTS
* COST
PROVIDE DATA TO ODW IN SUPPORT
OF REGULATORY PROCESSES
-------
MAJOR POLLUTANTS ASSOCIATED WITH
ROW CROP AGRICULTURE
1. SEDIMENTS
2. PHOSPHORUS
3. NITRATES
4. PESTICIDES
-------
U.S. Environmental Protection Agency
Drinking Water Research Division
Heidelberg College
Water Quality Laboratory
Bowling Green Water Treatment Plant
Fremont Water Treatment Plant
Ohio American Water Co.
Tiffin District
-------
Lakt Erie
OHIO
I
SANDUSKY RIVER BASIN
Figure 2. Location of the Sandusky River Basin, the Sandusky River, and Tiffin Ohio.
100
-------
PRO IFPT PF^TiriRFQ
rnwJCw I r CO I IvIUCO
alachlor
atrazine
de-ethyl atrazine
de-isopropyl atrazine
metolachlor
carbofuran
butylate
cyanazine
metribuzin
chlorpyrifos
simazine
linuron
EPTC
diazinon
fonofos
terbufos
phorate
ethoprop
trifluralin
penoxalin
-------
Table 4. Comparison of peak herbicide concentrations in tap water of Tiffn,
Fremont, and Bowling Green, Ohio for 1983 with previous measurements in Tiffin
and maximum values reported in the United States in 1977.
Tiffin
Fremont
Bowling Green
Tiffin
Max. Obs.
Tap Water
Tap Water
Tap Water
Tap Water
Cone.
1983
1983
1983
1980-82
1977*
ug/L
ug/L
ug/L
ug/L
ug/L
Simazine
Atrazine
Alachlor
Metolachlor
Linuron
Cyanazine
~Taken from National Research Council, 1977, Drinking Water and Health
0.63 0.13 0.35 1.9 detected
7.64 1.22 5.2 30 5.1
2.73 0.47 5.91 14.3 2.9
13.65 1.33 4.75 24.2 no data
0.41 — 0.39 — no data
1.49 0.39 1.92 2.4 detected
-------
o
CO
DAYS 2 6
100
80
2
O 60
111
cc
40
20
% REMOVAL
REVERSE OSMOSIS
FT-30
TIFFIN (6-12 TO 7-30 OF1966)
12
18
24
30
36
ALACHLOR ATRAZINE CYANIZINE METRIBUZEN
~ A O
METOLACHLOR LINURON SIMAZINE
J I I I l I I I L
j L
42
J I I l I I ¦ I I I I I I I I I I I I I I I I I I l L
100 200 300 400 500 600 700 800
HOURS RUNNING
900 1,000
-------
% REMOVAL
REVERSE OSMOSIS
CELLULOSE ACETATE
TIFFIN (6-12 TO 7-30 OF 1968)
DAYS 2 6
100
80
o
60
o
2
LU
DC 40
20
12
18
24
30
36
42
ALACHLOR ATRA23NE METOLACHLOR
~ A 0
100 200 300 400 500 600 700 800 900 1,000
HOURS RUNNING
-------
% REMOVAL
REVERSE OSMOSIS
NYLON AMIDE
TIFFIN (6-12 TO 7-30 OF 1908)
DAYS 2 6
100
80
o
cn
O 60
2
UJ
8?
20
ALACHLOR
~
ATRAZINE CYANIZINE METRIBUZEN
A 0
METOLACHLOR UNURON SIMAZINE
¦ A m
_L
l
100 200 300 400 500 600 700
HOURS RUNNING
800
900 1,000
-------
REVERSE OSMOSIS
FT-30, % REMOVAL
ALACHLOR
IN-HOU8I
1441
ALACHLOM
' 1 i i i i ¦ i ¦ i ¦ i . i
t 20 40 00 00 100 120 140 100 100 200
HOURS RUNNING
40
30
20
10
0
-as
-30
00
00
120
0
20
40
100
140
100
100
200
HOURS RUNNING
too
PBUL CONC.
140
1«
1»
IM
1M
M
HOURS RUNNING
106
-------
i
o
s
a.
REVERSE OSMOSIS
DESAL1
LINURON
IN-HOUSE, % REMOVAL
8-13-87
50 100 150 200 250 300 350 400 450 500 550 600 650
HOURS RUNNING
3 140
tfj 120
g 100
80
60
40
i
20
50 100 150 200 250 300 350 400 450 500 550 600 650
HOURS RUNNING
MP. KML CONC.
OJr-4"
m+Jm
-Up.
50 100 150 200 250 300 350 400 450 500 550 600 650
HOURS RUNNING
107
-------
REVERSE OSMOSIS
NYLON AMIDE
CARBOFURAN
IN-HOUSE, % REMOVAL
9-15-87
CARBOFURAN
600
800
200
400
1,000 1,200 1,400 1,600
HOURS RUNNING
CARBOFURAN
0 200 400 600 800 1,000 1,200 1,400 1,600
HOURS RUNNING
1,000 1,200 1,400 1,600
400
600
800
200
HOURS RUNNING
108
-------
REVERSE OSMOSIS
NYLON AMIDE
SIMAZINE
IN-HOUSE, % REMOVAL
9-15-87
100
SIMAZINE
200
400
600
800
1,000 1,200 1,400 1,600
HOURS RUNNING
•20
-40
600
800
200
400
1,000 1,200 1,400 1,600
HOURS RUNNING
160
140
120 —ys.
100
INF. PERM. CONC.
1,000 1,200 1,400 1,600
800
400
600
200
HOURS RUNNING
109
-------
Percent Removal of Cyanazine, Unuron, Alachlor, and
Metolachlor by a variety of Reverse Osmosis Membranes
100
90
80
70
60
50
IE 40
££ 30
20
10
0
Tiffin, Oh
Tiffin, Oh
Tiffin, Oh
Tiffin, Oh In-Houi
o
2
in
In-Hou
I
ii
CYANAZINE UNURON
* REMOVAL FLUCTUATED
ALACHLOR
METOLACHLOR
¦ CELLULOSE ACETATE
~ THIN FILM FT-30
POLY AMIDE
«m THIN FILM DSI
-------
Percent Removal of Simazine, Metribuzin and
Atrazine by a variety of Reverse Osmosis Membranes
100
90
80
70
60
50
40
30
20
10
0
TMjjln, Oh jWHou*eTlflln, Oh
JJ
ii.
Tiffin, Oh In-Hous
*
1
SIMAZINE
METRIBUZIN
ATRAZINE
* REMOVAL FLUCTUATED
¦ CELLULOSE ACETATE
~ THIN FILM FT-30
POLY AMIDE
THIN FILM DSI
-------
REGENERATE
DISCHARGE
PERMEATE
REJECT WATER
ION EXCHANGE RESIN
WELL
FILTER
STORAGE
TANK
STORAGE
TANK
GAC
IER
GAC
GAC
REVERSE
OSMOSIS
FLOW SCHEMATIC FOR
PILOT PLANT - SUFFOLK COUNTY, NEW YORK
-------
WATER QUALITY OF SELECTED
WELL
CONSTITUENT RANGE AVERAGE
ALDICARB SULFOXIDE 32-59 ug/L 40 ug/L
ALDICARB SULFONE 40-57 ug/L 47 ug/L
CARBOFURAN 13-27 ug/L 15 ug/L
1,2-DICHLOROPROPANE 19-33 ug/L 24 ug/L
1,2,3-TRICHLOROPROPANE 8-20 ug/L 13 ug/L
NITRATE 9.8-11.7 ug/L 10.4 ug/L
-------
SUMMARY OF REVERSE OSMOSIS PILOT PLANT DATA *
MONTH
ALDICARB SULFONE
ALDICARB SULFOXIDE
R
P
C
R
P
C
JULY
31.4
1.3
79.0
-
-
-
AUG.
27.2
1.2
79.7
20.0
<1
58.2
SEPT.
22.8
1.0
68.4
17.2
<1
51.4
OCT.
21.6
1.0
54.6
17.4
<1
42.3
NOV.
20.3
1.0
45.8
16.2
<1
36.0
DEC.
19.0
1.0
42.3
14.7
<1
32.7
JAN.
17.4
1.5
42.2
14.0
<1
34.9
FEB.
16.5
1.0
42.5
13.4
<1
33.8
MAR.
16.3
1.0
43.1
13.0
<1
33.2
APRIL
14.8
1.0
42.8
13.0
<1
34.3
MAY
14.7
1.0
42.0
11.6
<1
32.8
JUNE
14.0
1.0
39.6
11.2
<1
30.5
R = RAW
P = PERMEATE
C = CONCENTRATE
* = MONTHLY AVERAGES (ug/L)
-------
SUMMARY OF REVERSE OSMOSIS PILOT PLANT DATA *
MONTH
1,2-DICHLOROPROPANE
CARBOFURAN
R
P
C
R
P
C
JULY
-
-
-
9.8
<1
24.5
AUG.
22.2
6.5
43.0
8.0
<1
21.5
SEPT.
22.2
6.4
41.0
8.1
<1
22.5
OCT.
20.6
6.8
49.6
7.8
<1
19.5
NOV.
18.2
6.4
44.0
6.2
<1
13.0
DEC.
19.0
5.8
35.7
5.8
<1
13.8
JAN.
20.2
8.5
42.7
5.9
<1
14.0
FEB.
21.0
6.5
45.3
5.7
<1
13.4
MAR.
19.5
6.3
31.3
5.3
<1
12.8
APRIL
20.0
7.5
46.3
4.9
<1
12.6
MAY
17.5
8.3
47.8
4.7
<1
13.3
JUNE
18.8
6.9
36.7
4.3
<1
12.7
R = RAW
P = PERMEATE
C = CONCENTRATE
* - MONTHLY AVERAGES (ug/L)
-------
REVERSE OSMOSIS
AVERAGE PERCENT REMOVAL
SUFFOLK COUNTY, NY.
DUPONT POLYAMIDE HOLLOW FIBER
PESTICIDE CONTAMINATED GROUNDWATER
I
LU
o
a:
LU
Q.
100
80
UJ 60
CC
40
20
CARBOFURAN
ALDICARB ALDICARB
SULFONE SULFOXIDE
1,2-DCP
* SkWhs running
-------
LABORATORY RO DATA (POU)
Be
0.043
>97.7
Hg(l)
0.017
>97.1
Pb
0.28
>98.3
F
5.95
98.3
Cd
0.045
>95.6
U( total)
69.2ug/L
>99.0
U(total)
182.5 ug/L
>99.0
As(+3)
0.101
73.3
-------
REVERSE OSMOSIS TREATMENT
DISADVANTAGES
* HIGH ENERGY USER - HIGH COSTS
OF CONTAMINANTS
* PRETREATMENT REQUIREMENTS
* 20-50 PERCENT WASTE WATER
-------
REVERSE OSMOSIS TREATMENT
ADVANTAGES
* EFFECTIVE ON A WIDE VARIETY
OF CONTAMINANTS
* SMALL SPACE REQUIREMENTS
* PACKAGE PLANT SYSTEM
-------
CONCLUSIONS
1. REVERSE OSMOSIS REMOVES LOW MOLECULAR
WEIGHT VOLATILE ORGANIC CONTAMINANTS AND
CERTAIN PESTICIDES.
2. REMOVALS VARY ACCORDING TO TYPE OF
POLYMER MEMBRANE USED.
3. REVERSE OSMOSIS SHOWS PROMISE FOR DUAL
PROBLEMS SUCH AS INORGANIC/ORGANIC
CONTAMINATION.
4. REVERSE OSMOSIS SHOWS PROMISE FOR HOME
USAGE, SMALL UTILITIES. AND IN PLACE OF
SEVERAL PROCESSES AT LARGE UTILITIES.
-------
CO
CO
o
CO
o
w
CO
w
>
w
g £
2o
E-!
a o
Oh
Pu ^
< ^
£ g
S w
J3 Pi
121
-------
Questions and Answers for Carol Fronk
Elizabeth Kawczynski - When will the concentrate disposal option start and how
long do you plan to run those experiments?
Carol - Right now we are just running a single pass test and basically the concen-
trate simply goes to waste. At present, I'm in negotiations with several manufac-
turers for an eight membrane unit that we will begin testing this winter. I would
also like to take a look at pervaporation membranes in two different ways: 1) as
a stand alone technology, and 2) as a concentrate cleanup technology for the
multiple pass unit. There are many research avenues that we can look at but I
believe a technology that can pull just VOC molecules out would be a very good
step. I will be looking at concentrations for the influent of 130 ug/L and for
the concentrate 130 ug/L and up.
Bob Bergman - When you say multiple pass are you referring to concentrate staging
or product water staging? Are you taking the product out of one and going to
another unit or are you just talking about staging the concentrate?
Carol - Well, I will be taking the product (permeate) out of one and putting it
into another stage. For example if I get 90% removal of benzene, starting out at
100 micrograms per liter and getting down to 10, I will want to further reduce that
from a regulatory standpoint. Most of the maximum concentration levels (HCL's)
are going to be set at approximately 5 micrograms per liter, so that's the level
I will shoot for.
Dave Paulson - You mention R0 and UF, have you considered nanofiltration, this
new class of membranes, in fact some really new membranes as a possibility.
Carol - For the VOC's and pesticides?
Dave Paulson - Yes, especially pesticides.
Carol - If you are looking at lower molecular weight compounds below 100, say the
VOC's, I have a dickens of a time just finding reverse osmosis membranes which
work. They do so because of the chemical and physical filtration and the chemical
diffusion.
Dave Paulson - I agree it has to be a large molecule, I'm not sure if aldicarb
pesticides would work.
Carol - Yes, I plan to look at nano and ultrafiltration for pesticides this coming
year, that is another phase.
Dave Paulson - In an editorial comment, I think its a little dangerous when you
list the membranes out to use the polymeric family like polyamide and then sort
of contrast with thin-film composite when in fact a lot of those composites can
be considered polyamide too, I think the DuPont polyamide you talked about would
be considered a homogenous polyamide.
122
-------
Carol - Well that's something that I think is extremely important. I had a very
difficult time getting any information on what the polymers were. When I started
I got a lot of "its proprietary" from the manufacterers, which is great but it
makes it very difficult then to determine and correlate why we get a certain
percent removal. I realize that the manufacturers are protecting their market and
its business, but nevertheless it is difficult to determine why a certain membrane
outperforms another.
Dave Paulson - It's in the literature now, it's in all the membrane conference
papers, the major membranes that you address there, their chemical structure in
the literature. Anyway they need to be differentiated but I think it's dangerous
to say TFC's work and polyamide doesn't. That's my point.
Carol - Host of the thin-film composites by in large are superior to the others,
on certain classes of compounds. As I said before for the pesticides, all the
membranes including the cellulose acetate and the DuPont polyamide remove them
well.
Dave Paulson - I agree that they are better, it's not necessarily because of the
different chemistry.
Carol - Well maybe it is because of how they were formulated, for example how
they're heated, but there is something that is different about each of them. I
think that's important, so that people do not go out and just say that any RO
membrane will remove any compound, that's not true. You have to have a specific
membrane to remove a certain class of compounds.
123
-------
Dr. Mark Wiesner - Rice University
I will briefly summarize some of the research that I have been involved in
with membranes for the last five years and then give you an idea of some of the
current work that we are doing at Rice University as well as an indication of
where our work is headed. As sort of an introduction to the presentation, what
I've tried to illustrate here are a few of the issues this research addresses.
What I have tried to show here is a raw water being introduced onto a membrane
that contains colloids of various sizes and some molecular material that might be
humic or fulvic acid. And of course this material coming on here in the cross flow
configuration with the pressure applied across the membrane, can be deposited on
the surface of the membrane producing an additional layer of resistance to the flow
across the membrane and reducing membrane flux. In addition, if materials are
small enough to penetrate into the interior of the membrane, they can of course
pass through the membrane effecting the permeate quality or becoming deposited
within the interior of the membrane constricting membrane pores and again reducing
flux. To generalize, I guess you would say that decreases in flux associated with
deposition at the interior of the membrane are more irreversible than they are
reversible in nature, and the deposition on the surface of the membrane is more
reversible then it is irreversible.
We are trying to understand some the factors that control mass transport of
material up to and through membranes as it effects both membrane flux (or
productivity), as well as permeate quality. The past research that I will
summarize is the work of Veronique Turcaud who did her Masters and Doctorate at
the Lyonnaise des Eaux, in Le Pecq., France where I was her research advisor and
where Francois Fiessinger paid the bills. She was looking at two areas. One is
coagulation pretreatment for ultrafiltration, the idea being that we can somehow
take materials in the raw water and change the distribution of their sizes or
rejection characteristics vis-a-vis the membrane; both to remove materials that
might not be removed otherwise and, to improve membrane flux. The other thing
she was looking at, very much related, is the permeation behavior of various
colloids and macromolecules, that might represent fractions of materials found in
natural waters.
When I said fractions of material that you might find in natural waters you
do again need to use your imagination just a little bit. For example we are using
humic acid here, that's "humic acid in a jar". The Fluka humic acid is similar
to Aldrich humic acid, so it's not a natural aquatic humic acid. It has roughly
the same molecular wefght characteristics as many humic acids in surface waters.
Instead of a fulvic acid, we are using again another "jar" material, tannic acid,
with a molecular weight of about 1700. We were able to make various kinds of
particles by flocculating humic acid under different conditions and form what we
called "zone-two" particles. These were a very stable suspension of 0.2 micron
particles that might be the size of bacteria or clays. We also fed dispersions
of kaolin onto these membranes representing of course, clay material in water, and
flocculated material here having a size on the order of 100 microns since we were
looking at the effects of coagulation pretreatment . As well as the sizes of the
materials indicated here you have some indication of the transport of material in
membranes in general. You can think of these as order-of-magnitude estimates of
back transport velocity. Or, it may be better to think of these as mass transfer
coefficients. Brownian diffusion and a couple of hydrodynamic mechanisms, any
indication which might move materials away from the membrane. We would like to
compare this back-transport with convective transport to the membrane. Where our
124
-------
convective transport to the membrane is greater than the back-transport, materials
should be deposited on the membrane and create an additional layer of resistance
to permeate flux. Just as an example of the sort of experiment that we did,
Veronique would start up her system and establish an equilibrium flux with ultra-
pure water at the beginning. After about an hour the dispersion was introduced
onto the membrane (in this case tannic acid). In the case of tannic acid we saw
a very rapid initial drop in flux, followed by two subsequent stages of slower
reduction of flux. After a while we shut down the experiment and did a surface
wash across the membrane to remove material that might be deposited there, followed
by a chemical wash. That recuperated a certain fraction of the initial flux. But
there is a large fraction of the flux that we were not able to recuperate. I
should mention, these were hydrophobic membranes and so we tend attribute this
irreversible fouling to adsorption onto or within the membrane. For these
membranes, the problem of deposition on the surface of the membrane is something
like a 10 or 20% problem, were the problem of loss of membrane flux due to
adsorption on or in the membrane is more like a 40% problem. The material that
Mark Clark presented indicates that for hydrophilic membranes these portions might
be changed considerably. When we looked at the flux behavior for each on of these
materials (dispersions) our first qualitative analysis of our ideas about particle
transport in membranes were more or less substantiated. That is, flocculated
material tends not to be deposited on the membrane whereas humic acid and certainly
the "zone-two" particles were deposited on the membrane and caused quite a bit of
fouling.
You might come to the conclusion that since we were able to coagulate humic
and tannic acids really very well and since this material causes a lot of irrever-
sible fouling, it might be an interesting sort of pretreatment to add coagulants
to water and flocculate that material and end up more in the case of no deposition
on the membrane and very good membrane fluxes. The problem with that is that
natural waters are not nearly as cooperative as the fake stuff that we were using.
Shown here is a breakdown of the composition determined by pyrolysis GC/MS of the
DOC in Seine river water and the coagulated waters that Veronique was using. And
again, after coagulation what we were seeing for this water was that the dominant
fraction of DOC was present as polysaccaride material, which other work at the
Lyonnaise labs had indicated by direct speciation, deposited on the membrane as
causing the irreversible fouling of these membranes. Conversely, there was pretty
good removal of the polyhydroxide aromatics which tended to be larger molecular
weight material. The polysaccarides were smaller molecular weight materials.
In terms of the work we are doing currently at Rice University, I've divided
it into three different areas here. We are first and foremost trying to develop
models for process performance based on fundamental principles. Our primary
interest is in the area of particle transport and colloidal fouling. We would like
to get into more adsorption work in the future. A second area of concentration
here, involves the development of these models for process performance. We would
like to combine the models with some cost information on membranes and use those
in the context of non-linear optimization to look for favorable configurations for
membrane design and application. We are also involved in some applied work, doing
some field evaluations of membrane processes. The last thing I would like to do
in this presentation is to go into some detail on each one of these areas.
In terms of the development of these models for membrane performance, as I
mentioned, we are interested primarily, in the effect of colloid size and surface
chemistry on particle transport in porous channels and membranes. That sort of
125
-------
information in conjunction eventually with information on adsorption, will allow
us to develop models describing the overall performance of membranes and to use
those models to screen raw waters. There is another reason for looking at particle
transport and collidal fouling of the system, that has to do with the question "if
you are going to introduce a particle to a membrane unit what sort of a particle
would you like to introduce?" If you are going about doing some kind of
coagulation pretreatment then you should have a goal for that pretreatment and have
an idea about the sort of particle with respect to size and charge and so forth
that you will engineer to put onto the membrane. If you are using membranes from
a membrane reactor standpoint again using PAC on membranes, then you would like
to ask yourself "what is the size of PAC that I would like to put on the membrane."
You might have different criteria for particle transport and fouling then you would
have for favorable adsorption,and similarly, if ion exchange is used in conjunction
with membranes. Finally, particle transport may be important over the issue of
bio-fouling. If I idealize bacteria as particles, then the process of bio-fouling,
bacterial deposition, and subsequent development of bio-films will be effected by
the particle transport to membranes.
Our approach is to look at some of the forces that affect particle transport
in membranes. There has been a lot of work done on this, in different contexts.
George Bel fort has looked at hydrodynamic lift, we are looking at diffused layer
and London interactions, gravity force, so forth. These are two things that would
go into a particle trajectory analysis very easily. Then we also want to super-
impose on that things such as Brownian diffusion and shear-enhanced diffusion, to
come up with an overall model for the distribution of material of different sizes
as they flow over a membrane. To experimentally evaluate the model, we have con-
structed a filtration cell from which we can collect permeate from the membrane
and determine flux by weighing the permeate. We operate with a flow moving across
the membrane, and into that flow we inject our particles. When we make an injection
of particles, we observe the residence time distribution of the particles as they
come out the other side of the membrane channel. We can compare an observed resi-
dence time distribution with the distribution calculated based on the various
mechanisms of transport that I just mentioned. This Figure gives you a feel for
how data from those experiments looks, this is a case, an injection of .5 and 3
micron particles. We get two different peaks corresponding to each particle size,
and their superimposed residence time distributions. Most of our experiments in
fact have been with single sizes of particles and this Figure is a summary of some
of those experiments. I have plotted a kind of nondimensional mean particle velo-
city across the membrane as a function of the nondimensional permeation rate. We
get very different responses for different particle sizes.
The simulation/optimization work awaits the results of the process perfor-
mance models but we are beginning to do some cost modeling. I mentioned that our
goal is to take cost modeling information and combine it with the process perfor-
mance models, in the context non-linear optimization. This will allow us to sift
through some of the possible design configurations for membranes. Just as an
example, consider the fact that we have about 100 years of experience with rapid
sand or deep bed. Over that time, we have come to a sort of adhoc optimization
of the process. I think there is still a lot of work to be done there. But, I
would suggest that by comparison with deep bed filtration, membrane processes
present many more variables. We've got all the obvious normal noises; pressure
drop across the membrane, cross flow velocity, membrane pore size, configuration
of the membrane, and so forth. So when you add up all the permeations there are
a lot of ways you can design a membrane unit and optimization techniques should
help us identify some of the most promising designs. A related issue is that of
126
-------
the use of membrane processes in combinations; for example, microfiltration as
pretreatment for reverse osmosis, or nanofiltration. We would also like to
identify the effects of raw water characteristics on process selection. On the
other end of the pipe if you are designing a membrane process for particle removal
versus disinfection by-product removal or SOCs what would be the effect of a change
in the DBP regulations on the optimal membrane process design?
Finally, we would like to compare how membranes perform for given treatment
objectives with other combinations of processes such as conventional filtration,
GAC adsorption and so forth. To give you a feel of what the results of such an
analysis might look like, I'll show you some results from my Ph.D. work at John
Hopkins where we were asking the question "When would you use direct filtration,
contact filtration, or conventional treatment as a function of various raw water
quality parameters?" In this case the water quality parameters are particle
diameter and concentration of the raw water. We could add other axes as well, for
example TOC. We would like to find out where membranes stand in comparison with
these processes as well as others.
Since May we have been involved with some field evaluations of membranes.
We're looking at issues that are little different than some of the other pilot
studies that we have heard about today. There is a fair amount of experience with
hyper filtration and the work that has been presented here today on nanofiltration
and ultrafiltration, adds to our experience there as well. We are looking at
microfiltration and ultrafiltration. There is a very new inorganic ultrafiltration
membrane that just hit the market, and we are looking at that. So one difference
is that we are looking at ceramic membranes rather than polymeric membranes. We
are also interested in contrasting turbulent flow membranes versus laminar flow
units. A lot of the previous work has been done on ground waters. In our work on
surface waters we are treating an East Texas river water that is relatively high
in organics, (7 to 10 milligrams per liter TOC), and has high turbidities. We are
also looking at coagulation pretreatment.
I will briefly present some of the results of that pilot work. We've looked
so far at three different membranes; a 0.8 micron membrane, a 0.2 and 0.05. The
0.05 micron membrane some might argue is a microfiltration membrane. For the 0.05
micron membrane we are operating at GFDs of around 300. We are also comparing how
these membranes perform with conventional treatment. We're operating this pilot
system out at the Houston Water Treatment Plant, so we can sample a battery of
filters, and compare water from the conventional plant with our pilot system. The
conventional plant is similar for removal of DOC, perhaps a little higher, than
our membrane unit. The particle removals are also very similar and THMFP removals
are similar. The turbidity however, is much lower on the membrane unit. We are
measuring bacterial removal which is very good overall for both the conventional
and the ultrafiltration module here. We're not set up to do viruses, but we are
starting to think about how we can do that. Just as a comment, the removals
observed here between conventional treatment and membrane filtration, are not
particularly surprising in light of the fact that if you look at TOC removal across
a conventional treatment plant I think the US average is around 40%.
127
-------
Question and Answers for Mark Wiesner
Elizabeth Kawczynski - The last research project did you say you are using Houston
water at the water treatment plant?
Mark - Actually we have two raw waters, Trinity river water and San Jacinto River
water. So that was fun when the Trinity river flooded and we got some very
different water qualities.
Elizabeth Kawczynski - What kind of particle counter are you using?
Mark - Well we are characterizing particles two ways, the routine analyses that
we are doing are with a Coulter counter type of instrument. We start at 19 micron
apperture.
Mark Clark - Your particle elution results show that the small particle came out
first, which is what you predicted; right? Wasn't that the left arm of your graph?
Mark - The small particle came out first that is correct.
Mark Clark - They were more susceptible to motion weren't they?
Mark - I'm not sure that we are really interpreting that data correctly. It was
just really over the last three weeks that the system has been debugged and ready
to start to producing data, so we really need to do some good thinking about our
results.
Joe Jacangelo - On the membrane where you were getting 300 gfd, is that run in a
recirculation mode?
Mark - Yes and no. Actually it is worse than it sounds, initial work was at 100%
recovery, essentially a recirculation mode without any wastage. We started there
and backed off to 95-90% recovery, and we will keeping working down.
Joe Jacangelo - What kind of recirculation rates were you using?
Mark - I can't remember the cross flow velocity. Just from a cost standpoint, I
know you had originally looked at ceramic membranes and compared costs with some
other membranes. Our fluxes would reduce the cost that you had originally
calculated. We are working at a delta P of about 50 psi so that will bring the
cost down in line, I think, with the polymeric membranes. Obviously the initial
capital cost of these things is still much higher than the polymeric.
Joe Jacangelo - Is that with pretreatment?
Mark - Those fluxes, yes. I don't know if you noticed, but without pretreatment
we were getting such low fluxes that without pretreatment it just didn't seem to
make sense, maybe we should look at that some more.
128
-------
Elizabeth Kawczynski - American Water Works Association Research Foundation
The American Water Works Association Research Foundation is a non-profit
organization that is dedicated to the implementation of a research effort to help
local utilities respond to regulatory requirements and high-priority concerns of
the industry. The research agenda is developed through a process of grass-roots
consultation with members, utility subscribers, and working professionals. Under
the umbrella of the five year plan, the Research Advisory Council recommends
projects that the Board of Trustees selects for funding.
During the 1990 funding cycle, two separate membrane projects were selected
for funding. Each project has a $200,000 budget. The 1990 membrane projects are:
"Pre and Post Treatment Optimization of Ultrafiltration Membranes" and "Membrane
Concentrate Disposal". In 1989, the Research Foundation solicited proposals for
a project on the "Evaluation of Low-Pressure Membrane Filtration for Particle
Removal" and a contract was awarded to the Boise Water Corporation. In 1989, the
Foundation published an "Assessment of Potable Water Membrane Applications and
Research Needs" which was complied by Dr. James Taylor of the University of Central
Florida. Dr. Taylor's report provides information on potable water membrane plants
in the United States and recommendations for future research needs that are
necessary to utilize membrane technology in potable water treatment.
Let me tell you about the new projects in membrane technology. The pre and
post treatment project has been awarded to J. M. Montgomery, Consulting
Engineerings who will be working with Lyonnaise des Eaux, East Bay Municipal Water
District and Contra Costa Water District. The objectives of the project will be
to evaluate the impact of pre and post treatment on low pressure membranes to
reduce fouling and remove DBP precursor material. The operation, maintenance
requirements and costs for low pressure membrane treatment will also be evaluated.
The project will begin in November 1990 and be conducted over a 22 month period.
The membrane concentrate disposal project has been awarded to Mickely and
Associates in Boulder, CO. The researchers will be gathering information on
regulatory requirements for the disposal of membrane concentrates, data on the
characterization of these wastes from various types of membranes, and report on
options for disposal and waste minimization. The project just started in July
and will be two years in length.
Someone asked Joe earlier how they could receive publications on the work
being done at Boise and Contra Costa. The Boise project is a Research Foundation
project and will have a final published report that will be available from AWWA,
who acts as our publication group.
There is one other project I would like to tell you about-the Water Industry
Data Base. The Research Foundation with AWWA as our contractor just completed a
survey effort of the 643 utilities serving a population of more than 50,000 or said
another way the utilities that provide drinking water to over half of the
population of the United States. In the survey document, we included an area to
indicate future interests in the area of treatments. The results, I just received
from a phone call to Denver, indicate the nine utilities of the 475 who responded
have conducted some research in Reverse Osmosis, eleven of the respondents
indicated a future interest in RO and eight said they had an interest in
ultrafiltration. Although, the numbers don't sound impressive, you need to realize
that these are responses from large utilities.
129
-------
It has been a pleasure to report to you on the "coming attractions" in the
area of membrane technology and potable water treatment. There seems to be a
growing interest in the application of membranes in drinking water and as a
technology of the future, the Research Foundation will be conducting the projects
I've told you about today. Stay tuned.
130
-------
Questions and Answers for Elizabeth Kawczynski
Dave Paulson - The pre and post grant to JM Montgomery?
Elizabeth - Yes.
Dave Paulson - The concentration disposal option is just a paper study?
Elizabeth - That will be pretty much a paper study, that is being done by Mickley
and Associates, a small engineering group in Boulder, Colorado, who have done some
membrane work, primarily in the waste management area for electrical power plants.
One of the members of the firm constructed a software system for electrical power
plant waste minimization membranes and they are trying to see if, maybe, they can
get that from EPRI. EPRI is our sister organization on the electrical utilities
side.
Dave Paulson - There is no particular target date for any interim reports, just
in two years the final report will be out?
Elizabeth - The way we deal with things is that the researchers are required to
produce quarterly reports, in general those are not made public, if you will, but
are rather meant to be progress reports. The other thing that I should point out
is that in all our research projects we develop a three person PAC (Project
Advisory Committee), I felt like I was having a PAC meeting here because we have
Bill Conlon on our membrane concentrate disposal PAC, Jim Taylor's on our pre and
post membrane PAC, and Kim Fox is on another PAC that I have. We try to involve
in our PACs someone from the regulatory world, university community, and utility
so that we have a well balanced set of interest, if you will, represented, so that
the researcher can be guided more smoothly.
131
-------
Fred Pontius - American Water Works Association
This afternoon, I would like to mention some things that are happening within
AWWA that would be helpful for you. I am wearing several different hats. One is
the technical hat in terms of technical committees. I think most folks here are
familiar with AWWA. We accomplish our goals and mission through volunteer
committees and we are fortunate to have the chairmen of the three committees within
AWWA involved with membranes here. I asked each one of them to tell the group what
is happening in their committee, rather than try to ad lib for them.
First is the "Water Desalting and Reuse Committee" under the Resources
Division. Bob Bergman is the Chairman.
Bob Bergman - Our committee has been functioning for a number of years. I
have been Chairman for about three years. We have been sponsoring technical
sessions at the Annual Conference each year. A committee report was published
in last November's AWWA Journal on membranes. We are starting two planning
handbooks. One is called "Desalting Planning Handbook," another is "Water
Reuse Planning Handbook". These documents will take several years to produce.
They're intended to be general documents, not detail design manuals, useful
to water supply decision makers looking at various alternatives for other
sources of water. They will discuss what technologies are out there, the
state-of-the-art, how they compare between one another, and some cost
information. The other thing we are looking at right now is the AWWA policy
statement on using reclaimed water. I do not know where that is going to end,
but AWWA policy does not talk very much about indirect reuse as part of the
policy statement. It talks about direct reuse as something they don't see
in the future. We want to make a little bit stronger statement on indirect
reuse because a lot of places have been using indirect reuse for a number of
years.
Let's move over to research in the "Membrane Technology Research Committee."
Mark Wiesner is the Chairman of that group.
Mark Wiesner - We met for the first time a little over a year ago. The goal
of the Research Division committee is to produce a research needs document.
Normally, committees have three years to do that. We've produced a first
draft now which we are planning to submit to the Research Division for their
perusal by November. In addition, we just submitted to AWWARF possible
research areas that they could develop RFPs on. One research area is fouling.
A second one on just bio-fouling and a third one on multi-phase membrane
separation process. As for most of our activities right now we hope to be
done with the research needs document by next summer and move on to other
activities. One of the normal courses of evolution in the committees is to
develop a Sunday Seminar, but this is a lead into what Jim will say here,
there is actually already a Sunday Seminar in progress.
Jim would you like to take over with the "Membrane Processes Committee", Jim
Taylor is the Chairman of this committee.
132
-------
Jim Taylor - The "Membranes Processes Committee" is under the Water Quality
Division and as Mark noted is, along with the Research and Desalting
Committee, preparing a Membrane Conference. It's an international conference.
The call for papers has been sent forth and it will be held in Orlando,
Florida, March 10-13, 1991. The conference will have technical sessions,
exhibits, exhibitions and hopefully opportunity for some hands on experience.
It's designed for both technical development and education of the utility
industry as to the potential for membrane applications at various points in
the states. That's been the main committee activity in addition to a manual
of practice. Those are the two activities of the "Membrane Processes
Committee".
Thank you Jim. The activities of those three committees fall under what is
called the Technical and Professional Council. A1 Stevens serves on the council.
Within AWWA, there is another structural organization under the Water Utility
Council. The Utility Council's mission is to evaluate and interact with EPA on
regulations and legislation. When the question is asked what is AWWA's position
on something, that is the arena within which AWWA's position is crafted. I think
it might be helpful for me to explain how that system works. When a regulation,
like disinfection, disinfection by-product regulation, or Phase 5, or any of the
regulations begins, AWWA under the Water Utility Council establishes a committee
referred to as a Technical Advisory Workgroup (TAW). That group is not intended
to be a standing committee. The life of that committee exists only as long as the
work on the regulation is continuing, in principle. That committee as early as
possible will form and begin monitoring the activity on a regulation, it's
development, as early as the start action request all the way through straw-man
drafts, through proposal and formal comment period, that can take several years,
depending on the regulation.
Now in terms of the disinfection, disinfection by-product regulation the TAW
is chaired by Dick Moser, American Water Works Service Company, and we have under
contract Metropolitan Water District of Southern California to support that TAW.
Mike McGuire is under contract to be what we call the TAW manager to manage and
interact with that TAW. I am a TAW manager with certain TAWs assigned to
particular regulations and Tim Chinn of our Washington office is a TAW manager
assigned to certain other regulations. At the moment, in terms of BAT, I won't
bore you by reading it, but we do surprisingly enough have a statement on BAT.
Those of you that know AWWA know sometimes it is tough to come out with a statement
on anything, but A1 and I wanted you to see this, we have a statement on BAT.
Generally our comments on a particular process like membranes will be
regulation specific, so if AWWA was to take a stand on membranes it would be within
the context of the particular regulation. I think the information that was
presented this morning has shown, and from my experience and the experience of
others, the effectiveness of membrane processes is not in dispute. It's not a
fully understood process. There are certainly a number of questions regarding
probability for achieving particular water quality objectives. I don't suspect
that we will find ourselves for example, like we did a number of years ago with
GAC when it was first proposed and many water utilities strongly opposed GAC from
being recognized a BAT for SOCs. In terms of the disinfection regulations, our
position will be within the context of the proposed regulation.
133
-------
The Technical Advisory Workgroup follows the regulation through development.
Another important thing they do is gather technical data. This is separate from
doing original research studies. As many of you may know, about one year ago AWWA
embarked on a fund raising effort to raise money so that it can increase its
activity in the regulatory arena through what is known as the water industry
technical action fund. To date, we have raised a little over 2.2 million dollars.
Right now under contract we're probably spending $700,000.00. That fund will
continue through 1993. A specific ending date for that fund has not been
established right now, but I imagine it may continue on depending on the need, and
of course the willingness of the utilities to continue paying. The purpose of that
money is to be available as needed to fund particular projects, recommended by the
Technical Advisory Workgroups. We have a number of projects like I mentioned under
way. None at the moment are specifically related to membranes. I would like to
ask you what role we may play in seeing this technology advanced to the point of
it's fitting within a regulatory framework. That is the goal of that particular
activity. It's a fund to gather technical data to help make the transition from
research bench-scale studies and pilot studies to addressing some of the broader
issues of implementation. One of the things that the technical workgroups look
at pretty carefully is the effectiveness of the technology proposed for a parti-
cular regulation, and its cost. There are two aspects to the term BAT, one is BAT
using the context of a technology that is appropriate for meeting a regulation.
The second is a legal context. What does a utility have to do before it can apply
for a variance or what must be considered for an exemption? So we, in terms of
the TAW activities, look at both of those, not just the strict technical issues
but also the legal issues.
134
-------
Questions and Answer for Fred Pontius
Dave Paulson - Does the AWWA have any direct linkages with EPA, for instance EPA
personnel sitting on the committees, common committees or acting as advisors in
some of these areas?
Fred - There are a couple of different ways that occurs. One is within the
Technical and Professional Council context of the committees that are represented
here. Formal membership is possible and I believe that cross representation exists
in that particular area. When it moves over to the Technical Advisory Workgroup
area, where we're actively critiquing or commenting on a regulation, there is
coordination, not formal membership. The regulation managers, for the regulations
that I am directly responsible for I talk with frequently, I believe Steve is
reasonably informed as to what TAW is doing in terms of studies and that sort of
thing.
Dave Paulson - It's sounds like it is not formal? Do you have, for instance, an
identified individual from the EPA Drinking Water Division who sits on the
committee.
Fred - EPA membership is formal in the case of the Technical and Professional
Council committees, but not in the case of the TAWs. The TAWs are not designed
to be balanced committees in terms of representing all interests. That's tough
to swallow but that's the reality of the organization. We try to get a reasonable
mix. In the regulatory and legislative process, the Water Utility Council has the
primary responsibility of speaking to the organization.
135
-------
Steve Clark - Office of Drinking Water, Washington, D.C.
This facility here is run basically by the Office of Research and Development.
The Office of Drinking Water writes and develops regulations and then enforces
them. We do have a division here. I work in the division that develops
regulations. There's a State Programs Division that implements them, and there's
the division here, headed right now by A1 Stevens, that helps us at both ends with
technical studies, surveys, etc., called the Technical Support Division, which is
not to be confused with the Office of Research and Development. Prior to taking
this job, which I took about a year and half ago I was the Chief of the Technology
Section. I dealt with all areas of technology in setting drinking water standards,
that is, evaluating which technologies would get us good removals down to the
levels of health concern for any kind of contaminant. I had a staff that helped
me with that and with how much it would cost. Those are really the two basic
technical issues: you have a health-based criterion and you have the maximum
contaminant level which approaches that health-based number, taking cost and
technical feasibility into consideration.
I was very pleased this morning, when I listened to some of the presentations,
to learn that people were focusing on disinfection by-product precursor removal
and, at the same time, the role of membranes simultaneously removing the microbes
which we are interested in. Essentially, we are dealing with disinfection as the
requirement to prevent certain microbial diseases, such as viral diseases,
Giardisis, and other diseases from protozoans. But, at the same time, we are
trying to balance that need for a disinfectant or disinfection process, with the
need to lower human exposure to toxic chemicals, some which are carcinogens, some
chemicals known, some unknown. It is a very complex area. As I go through the
talk here, I discuss some of the indicator disinfection by-products that we are
interested in and how we're balancing that with the view of microbial safety which
is really the other half of this.
I am assuming that at least half the audience has very little experience in
dealing with the Office of Drinking Water setting standards. In fact, I mentioned
to a few people, we've had the opportunity to have a very long-term relationship
with, for instance, granular activated carbon manufacturers and people in that
field, but I feel as if the relationship with people in the membranes field is just
beginning. So, I am going to give you some very basic information on the Safe
Drinking Water Act. There is one part of the Safe Drinking Water Act that we need
to focus on and that is Best Available Technology. If we can identify some
contaminants that need to be regulated because of health concerns, we need to then
look at feasibility. What the law says is that we should look at feasibility in
terms of setting maximum contaminant levels which are the enforceable levels, using
the best technology, treatment techniques, and other means that the EPA
Administrator finds... after examination at field scale. So we want to have
technologies that have not only been demonstrated at laboratory scale, for
instance, as was shown this morning, but also at field scale. Now we modify this
a little bit by saying the technology itself needs to be demonstrated at field
scale; we don't have to check every virus versus a membrane system or every type
of chemical. But we get a general idea that this technology works at field scale
and then have laboratory bench-scale and pilot-scale data possibly to verify that
the specific chemicals behave as we would theorize they would behave in the field.
That kind of information is gathered for us mostly by the Office of Research and
Development, Bob Clark's Laboratory, the Risk Reduction Engineering Laboratory,
here in Cincinnati, but also to some extent by A1 Stevens' group and also by the
industry itself. The published literature is certainly available to us in our
deliberations on setting what is Best Available Technology. We utilize the
136
-------
literature, and we have contracts with consulting engineers. So, we put together
documents called Cost and Technology documents that give us an idea of what's
available and how much it costs and how well it works under various conditions.
I could talk a hour on just this topic but I just want to give you a general idea
of the context we are in. Finally, when we say we take cost into consideration,
we are looking at the cost of large metropolitan utilities -- places like San
Francisco, Boston, or Cincinnati, as opposed to very small systems which we also
regulate, down to those systems that serve 25 people, where the economies of scale
are so poor that our economic finding would be quite different if we focused on
those systems rather than the large systems in terms of what's affordable to the
consumer. So, as far as affordability and feasibility are concerned, we are
focusing on large metropolitan systems (about 50 to 100 million gallons per day),
places that can afford fairly sophisticated technologies and can operate them.
I think Carol Ann Fronk talked a little bit this morning about various
contaminant categories. I just want to mention that we are considering for
regulation a variety of inorganic contaminant chemicals and radionuclides. We
are at least proposing that membrane technologies, for instance, electrodialysis
and reverse osmosis, be considered Best Available Technologies for these kinds of
contaminants, for example, cadmium, chromium, radium, etc. For synthetic organic
chemicals, I don't think we have made any such proposal, but as you can see,
probably in a mixed contaminant situation, it might be feasible to simultaneously
remove nitrate and pesticides; that would be a very specialized application. What
I think we are most interested in, at least what I'm most interested in, in this
review and just in hearing some of the papers this morning, is the idea of using
membrane separation processes to get removal of viruses in particular. The virus
we are focusing on right now from a safety aspect is hepatitis A virus, but we have
also looked at other viruses that are much easier to measure. I think there are
only one or two scientists that can measure hepatitis A virus in drinking water
so that could be a limitation we realize.
Also, disinfection by-products—removal of precursors is probably the most
important area for application of membrane separation, but it is also an area we
are focusing on for setting the standards. Right now we have one disinfection by-
product group that's been regulated since 1979, the total trihalomethanes.that is,
chloroform and its bromonated analogues. This group is primarily considered a by-
product of chlorination which is the predominant disinfection process as you are
probably aware. We are also looking at setting standards for other disinfection
by-products in the future, and I will talk about that in a minute. First, I want
to just mention that I think a number of you are aware of this because I gathered
from your talks that you are interested in demonstrating that some of these
membrane technologies and separation technologies would meet these criteria.
Now these criteria are for the Surface Water Treatment Rule which is a final
rule that within the next year or so various-size water utilities will have to
start meeting. The criteria are based on either removal or inactivation of Giardia
organisms and viruses. I have already mentioned that we are focusing on hepatitis
A for the specific values. For inactivation, we are looking at CT values, that
is, concentration of disinfection over some period of time. Giardia is right now
the target organism. We have some concern with a new organism. I won't go into
any level of detail on Cryptosporidium which may be more resistant at least to
chlorination and therefore is more of a concern for systems that have somewhat poor
filtration or where there are a number of surface water systems, For instance,
New York City; Boston; and Portland, Oregon have no filtration and rely solely on
137
-------
disinfection with chlorine to obtain CT values and provide microbiologically safe
water. These Cryptosporidium organisms may be resistant enough to get through
that disinfection system and cause some level of disease. There are a lot of
questions about how infective these things are and how serious the disease is.
That question should be answered about the same time that the Disinfection By-
product Rule is promulgated. Maybe at the end I will give you a little idea on
what the schedule will be, because that's become very complicated. In the next
few years, we should have a rule for disinfection by-products, disinfectant levels
(that is, levels of chlorine, etc. at the tap), and ground-water disinfection which
is the second major bullet here.
The Ground-Water Disinfection Rule is primarily going to focus on some sort
of inactivation requirement for viruses in vulnerable ground waters. The second
little bullet underneath it is for variances for systems that don't need to
disinfect. If they can demonstrate that they are not susceptible to viral con-
tamination, they can continue not to disinfect. We also need to look at the
requirement for disinfection residual under the Surface Water Treatment Rule, the
last bullet here. Distribution system residuals are right now required for all
surface water systems. The systems have to maintain a distribution system residual
under certain requirements within the Surface Water Treatment Rule. We are still
debating that issue for the Ground Water Rule, so it may be, for instance, that
the membrane system demonstrated will remove 4 logs of viruses absolutely and the
system won't need that residual disinfectant. Then there is the question of safety
and reliability, etc. The idea is that we have these basic performance criteria
for the microbes rather than measurement of the microbes at the tap.
Controlling disinfection by-products, as I mentioned, is really a question
of balancing risk of exposure to microbial agents, which cause mostly acute
infections that generally people recover from, with this idea of exposure to
chemical toxicants, for instance, chloroform, which is demonstrated to be an animal
carcinogen. There is some concern that there are some measurable cancer risks due
to human exposure to these known animal carcinogens. I don't think any of them
at this point are known human carcinogens, but our concern is rather that, based
on extrapolation models, there are some human risks in the case of chlorination
by-products. We are dealing with a very large population, roughly 150-200 million
people exposed, so it's a fairly big question. Now, what are some of the chemicals
of concern? Most of the literature in this area focuses on the total
trihalomethanes, because they were discovered earlier than the others, somewhere
around the mid-1970s and regulated by EPA in 1979.
Subsequently we've done studies of other by-products of chlorination, and this
second group next to the trihalomethanes, or the TTHMs, the haloacetic acids,
seems to be the next in importance considering concentration and also possibly
toxicity. At this point, the preliminary indications are that at least two of the
haloacetic acids, di- and tri-chloroacetic acid, will probably be carcinogens, but
possibly also fairly potent neurotoxicants causing some nervous system damage at
very low concentrations. These chemicals tend to occur in chlorinated drinking
waters at relatively low pH levels, whereas trihalomethanes occur at higher pH
levels. So, we are trying to pick indicator groups that would be indicators of
overall safety taking into consideration things like pH, etc. So these two might
be the leading indicators of the chemical quality of chlorinated waters. There
are some others that we are considering, but these seem to be less important.
Considering the group here and the time, we won't go into some of the other more
specialized chemicals.
138
-------
If we chloraminate water, that is, add chlorine and ammonia in various sequen-
ces, we can get similar chemicals to chlorination as you would expect. We probably
get some other chemicals which we can't measure, for instance, inorganic chlora-
mines and organochloramine compounds, but we can't measure those. A compound that
we think we might be able to measure that seems to be significant in chloraminated
waters is cyanogen chloride. Unfortunately this chemical is fairly unstable and
we are having difficulties maintaining stable samples that you might need in a
regulatory monitoring program. We also do not have a whole lot of information on
its toxicity. It was a war gas agent in World War I, and there are some basic data
from that. It seems to be analogous to hydrogen cyanide, but we still do not know,
so we are trying to gather more toxicological data on this by-product of chlorami-
nation.
Chlorine dioxide, another alternative disinfectant we are looking at, has
specifically the two anions as by-products, possibly some other things, but at
least chlorite and chlorate. For ozone, we are looking primarily at aldehydes,
specifically formaldehyde, which is considered by EPA a carcinogen by inhalation
and possibly not a carcinogen by ingestion. That's a key issue for ferreting out
the status of ozonation at this point. Also, the second compound there, an
inorganic chemical bromate, can occur under some conditions of ozonation, taking
bromide ion and oxidizing it up to bromate. We obtained a copy of a translated
Japanese study which shows that bromate, under the conditions of the experiment,
was a fairly potent carcinogen for some rats, which would be a major concern. So,
bromate may be an important chemical in those waters that contain bromide ion, in
systems that ozonate or possibly use chlorine dioxide or other powerful oxidants.
Chlorine might oxidize it too; I just don't think it would result in as high a
concentration. Finally, we are going to regulate the disinfectant residuals
themselves, that is set, maximum contaminant levels for chlorine, chloramines, and
chlorine dioxide. They would primarily be health-based, but again we need to
balance that with the idea of the necessity of maintaining a disinfectant residual.
As far as setting the maximum contaminant levels is concerned, we have three
basic approaches. I think the approach here that we need to talk about the most
is the first, that is, precursor removal--removing humic and fulvic acids, amino
acids, proteins, sugars, whatever is occurring in the raw water that's reacting
with the oxidant, for instance chlorine, in forming the disinfection by-products,
for instance the total trihalomethanes. Alternate oxidants may be a site-specific
circumstance at this point now that we are moving into regulating by-products of
the major oxidants, whereas formerly, using disinfectants like ozone plus chlora-
mines would produce very low trihalomethanes, and trihalomethanes were the only
things being regulated. At this point, we may be regulating formaldehyde and
bromate which could be a concern for certain waters that are ozonated and
chloraminated. For by-product removal, there are a variety of processes-
adsorption, reduction, possibly even biological treatment for instance, for
aldehydes using a fixed biological bed.
Probably more important for this talk and later discussion is how are we going
to set the maximum contaminant levels, and how are we going to determine what is
the Best Available Technology for setting the MCL? Let's take trihalomethanes as
an example. The total trihalomethanes standard now is 100 micrograms per liter.
It was set under somewhat different legal criteria. First of all, it is an interim
standard. It was a different section of the original law that has to do with
interim standards. It's a less stringent standard. Also, there have been changes
in the law. For instance, it made granular activated carbon a Best Available
Technology for these kinds of chemicals, the synthetic organic chemicals, and it
139
-------
has a somewhat changed definition of Best Available Technology itself. So the
approach that we are going to use, at least the approach we are thinking of using
right now, and of course this is tentative, is to define a surface water, at least
initially, in terms of total organic carbon and other parameters, such as pH,
alkalinity, etc. We are already in the process of doing that. Then, to define
some treatment system that would be applicable to a large surface water plant, for
instance, conventional treatment, possibly followed by granular activated carbon
adsorption for precursor removal. A system designed to remove turbidity plus do
some level of precursor removal is an example: targeting say 50% removal of total
organic carbon and then applying prechlorination to attain that 3 log or however
many logs is required to meet the requirements of the Surface Water Treatment Rule:
about a half a log inactivation of Giardia. because we've already removed some of
the Giardia in the filtration process and three or four logs of the viruses that
have gone through the filtration process. So at that point, we have disinfected
the water and we have created some by-products which we can measure and extrapolate
etc.
Then, we also need to maintain a disinfectant residual. For this model system
we've looked at free-chlorine. The reason we are looking at free-chlorine is it's
still considered the primary standard for disinfection; the disinfectant that
everything else is compared to. We are setting up our model system to later make
comparisons to ozonation, chloramination, etc. So, we have a system now that has
disinfected its water to kill the pathogenic microbes that are in the water and
have possibly added some more disinfectant to maintain a distribution system
residual in this really large metropolitan system by extrapolation, modeling, or
whatever. We can now come up with a tentative range of numbers for the total
trihalomethane standard. Once we've done that, we've identified the Best Available
Technology that I showed you in the very first slide, the Best Available Technology
that determines what the maximum contaminant level will be. Then the approach
would be up to individual water utilities to select technologies amoung those
1 isted for precursor removal: conventional, coagulation, granular activated carbon
adsorption, or membrane separation to meet that maximum contaminant level. One
job that we would then have left would be to list those technologies that are
considered Best Available Technologies for the purpose of granting variances. That
is, if a water system applies one of these technologies and still cannot meet the
MCLs for some reason, then it could be granted a waiver to the regulation, at least
temporarily, because it has applied Best Available Technology. This is where all
these other technologies fit in. They would be listed by EPA as technologies that
would be useful for granting variances rather than up front as technologies that
we have used to determine what the standard should be. The standards will probably
be based on a conventional technology, with the addition of carbon adsorption,
because carbon adsorption has been defined as a Best Available Technology for
synthetic organic chemicals by law.
So, we see membranes at this point fitting into the analysis in the phase
where people are complying with the disinfection by-products regulation, and also
complying with the disinfection regulation. As we've seen here this morning, we
are not only removing precursors in some instances but, in most instances, we are
removing Giardia and viruses which are targeted organisms and we are getting fairly
high removal rates. Because we get these high removal rates, we can have much
lower levels of disinfection, obviously giving us less exposure to disinfection
by-products. Lower levels of disinfection by-products would result from two
angles: one, lower disinfectant dose; the second, the removal of the precursors.
140
-------
One of the disadvantages as we see them, possibly, and we're just learning about
this area to be quite honest, is that pretreatment is a big unknown, particularly
for surface waters. We are dealing with very large systems using surface waters
of various qualities, for instance, the Ohio River here for Cincinnati. Would it
be feasible to treat the Ohio River for Cincinnati with membranes? They have a
plant handling about 200 million gallons per day. Is this a feasible technology
considering that water quality for removing Giardia virus and/or the disinfection
by-products? That's really a question that we need to know the answer to as a
regulatory agency. Although somewhat hypothetical, it's still a fairly important
question to help guide our thoughts and possibly even our research. And finally,
another disadvantage is waste disposal--partly due to some EPA regulations that
have to do with waste disposal and hazardous waste definitions, etc., that are
somewhat outside the purview of the Office of Drinking Water. Although we
understand them and we work with those folks, waste disposal is certainly a
concern.
Best Available Technology: we promulgated the Surface Water Treatment Rule
that really doesn't fall under the definition of Best Available Technology but
defines by performance criteria what technologies are applicable, that is, 4-log
removal of Giardia. or 4-log removal of viruses and 3-1og removal of Giardia. We
proposed that for certain inorganic chemicals, Best Available Technology is reverse
osmosis. By early next year, we should propose standards for radionuclides,
particularly radium and uranium. Right now, as it stands officially, we are going
to propose a Disinfection By-Products Rule next fall. That was until about three
or four weeks ago when a citizens group in Oregon sued the Agency, and told us we
had missed the statutory deadline for 25 chemicals, the disinfection by-products
being among those 25 chemicals. We are in the process of responding to the Federal
District Court in Washington telling them what our schedule should be. Once we're
on a court-ordered deadline, it makes the situation a little bit different, so we
need to go back and think about our research needs and about our analytical needs
as far as policy analysis is concerned, and give the court a more definite answer
than this deadline we've set for ourselves of late next year. The Ground-Water
Treatment Rule is in the same ballpark, and we are hoping to attach these two
rules.
141
-------
BEST AVAILABLE TECHNOLOGY
...feasible with the use of the best
technology, treatment techniques and
other means which the Administrator
finds, after examination for efficacy
under field conditions and not solely
under laboratory conditions, are
available (taking cost into
consideration).
142
-------
MAJOR CONTAMINANT CATEGORIES
o INORGANIC CHEMICALS
o RADIONUCLIDES
o SYNTHETIC ORGANIC CHEMICALS
o MICROBES
o DISINFECTION BY-PRODUCTS
143
-------
MICROBIOLOGICAL PERFORMANCE
CRITERIA
o SURFACE WATER
- 99.9% Removal/lnactivation of
Giardia
- 99.99% Removal/lnactivation of
viruses
o GROUND WATER
- 99.99% Removal/lnactivation of
viruses
- Variances
o DISTRIBUTION SYSTEM RESIDUALS
-------
DISINFECTION BY-PRODUCTS
o BALANCING MICROBIAL vs. CHEMICAL
RISKS
o CHEMICALS OF CONCERN
- TTHMs, Haloacetic Acids
- Cyanogen Chloride
- Chlorite, Chlorate
- Formaldehyde, Bromate
- Disinfectant Residuals
145
-------
DISINFECTION BY-PRODUCTS -
TREATMENT
o APPROACHES
- Precursor Removal
- Alternate Oxidants
- By-Product Removal
o MODEL SYSTEM APPROACH
- Surface Water (TOC)
- Treatment (Precursor Removal, Cl2)
- Distribution (Large, Cl2)
o PRECURSOR REMOVAL
- Conventional
- Granular Activated Carbon
- Membranes
146
-------
MEMBRANES FOR DISINFECTION
BY-PRODUCTS
0 ADVANTAGES
- High Removal Rates
- Decreased Disinfectant Doses
o DISADVANTAGES
- Pretreatment (Surface Water)
- Waste Disposal
147
-------
BEST AVAILABLE TECHNOLOGY
o PROMULGATED
- Surface Water Treatment
o PROPOSED
- Inorganics (Phase II)
o PENDING
- Radionuclides
- Disinfection By-Products
- Ground-Water Treatment
148
-------
Questions and Answers for Steve Clark
Bill Conlon - One of the things that I thought you should know in declaring
membranes BAT for radionuclides is that I was involved in running several
demonstration projects and we found that while running mass balances, that
cellulosic acetate membranes appeared to take up some of the radionuclides and the
membranes become hot eventually. We did not find that to be true with thin-film
composite membranes. Just thought I would pass that along.
Steve - That is a very important concern for the workers and we are concerned about
that.
Bill Conlon - The other thing is that I serve on several concentrate disposal
related committees and one of the issues that just recently came up too in Florida
is that a local DER office reported a conversation which occurred at a recent
underground injection committee meeting at which there was an EPA official from
Washington, the UIC group, and there was apparently talk about designating membrane
concentrate a hazardous waste rather than an industrial waste. We in the industry
feel that the concentrate really shouldn't even be an industrial waste let alone
a hazardous waste. Now if EPA on the one hand is forming regulations and rules
that determine membranes as BAT for different constituents, why then are you not
communicating with these folks within EPA who are making rumblings about passing
regulations that will make it impractical for the industry to use membrane
technology?
Steve - Let me start with the organization. I didn't mention this, but the Office
of Drinking Water contains a component that regulates underground injection which
is part of the same office that I work for. We do talk to them and our boss is
the same. Mike Cook, who directs the development of drinking water standards, is
the same guy who directs the development of standards for underground injection.
Now, to date, my understanding of what a hazardous waste is: something that fails
certain extraction procedure tests which are changing all the time, and I won't
even quote which test it has to be. If you take the waste and you run it though
the test as defined in the Federal Register, and it passes the test, then it's not
a hazardous waste. Now the designation of industrial waste, I believe, is made
mostly by State and local officials for pretreatment regulations.
Bill Conlon - The State of Florida exfacto concentrate committee that I serve on
is having another meeting of the committee on the 27th if anybody is interested,
in Tallahassee in Room 132 at 3:00 P.M. The contractor that is going to be
performing the concentrate disposal work for the AWWA Research Foundation will be
there along with other consultants that may want to attend, including industry
representatives. The main theme is to attempt to keep these rules and regulations
pragmatic, but one of the concerns of everyone is the feeling that EPA is going
to proclaim concentrate a hazardous waste. In Florida, initially for example,
there were only two categories to place the concentrate disposal permit and back
in the early days these were domestic waste or industrial waste. So, the
regulatory agency said well it doesn't fit the category domestic waste because it
doesn't contain fecal matter, so, we'll require an industrial waste permit. That's
when it began, and then when the NPDES regulations came out, I think, when they
recognized the pH, (because mostly back in those days we used cellulose acetate
membranes and the system had low pH feedwater because of the addition of acid to
prevent hydro!ization of the membrane). EPA then declared since you have a low
pH in your concentrate, you are, therefore, classified an industrial waste. So
that has stuck and now when we are dealing with deep well injection in Florida,
149
-------
and with the underground injection people they require tubing and packing, and they
require extra monitoring wells because they consider concentrate an industrial
waste. Well if it's just the pH EPA is concerned with, all we should have to do
is simply pretreat the concentrate by adjusting the pH before we dispose of it.
Then, it's no longer an industrial waste; or pretreat the concentrate in some other
way if it's another parameter causing the concentrate to be in the industrial waste
category for some reason. I think we need to remove the industrial waste misnomer
off of concentrate, because most concentrates that I have dealt with, the LSI is
on the plus side or positive,and not even an aggressive solution.
Steve - That's something I have to look into. I'm not all that familiar with those
programs. I think a lot of water waste would actually get a permit under the
industrial system, because of the types of effluents--alum, for example. It
depends on the State.
Fred Pontius - It depends on the State. If they fail the TCLP and they are
designated a hazardous waste, they go on to RCRA. If they don't, then it really
depends on what State, how the State wants to regulate it.
Bill Conlon - Well, you are talking about toxicity.
Fred Pontius - If it's a nonhazardous waste, it depends on how the State decides
how it wants to handle it. Different situations exist throughout the country.
Bill Conlon - Yes and I believe that Florida has the toughest water/wastewater
regulations but in terms of toxicity, there have been several cases reported where
and agency required the use of a fresh water minnow in the bioassay test to check
water being discharged into a intracoastal waterway, where the brackish water would
kill the fish alone without the concentrate being added. These are some of the
things that are reported to be happening that I think you should be aware of.
Steve - On underground injection, we are right there. On discharges to surface
water we are in the same major office. For determination of hazardous waste, it
is called the Office of Solid Waste. It's as different as the Office of Research
and Development; it's the major offices that are different. They have written to
us numerous times and have said here are the tests in the Federal Register that
you run. I've never seen anyone (people come to me all the time and say this is
a hazardous waste, this isn't) who came to me and said this is a hazardous waste
and here are the test results. Someone just makes a decision that it's a hazardous
waste without the test and that's not what that program is telling us in writing.
If you fail the test, then you're a hazardous waste.
Bill Conlon - If you spoke to Howard Rhodes, who is in the Office of FDER in
Tallahassee, he could explain to you the situation that's been occurring recently,
which has all of us "afraid of what is going to happen down the road".
Steve - But that's the difference between domestic and industrial versus hazardous
wastes. There are three different definitions.
Elizabeth Kawczynski - Did I understand you to say that in taking cost into
consideration to define Best Available Technology that cost consideration is made
for the larger systems? What's the cutoff for large systems?
150
-------
Steve - For determining what the maximum contaminant level will be, the primary
determination is made for large systems that are serving 50 to 100 thousand people
or more. For the entire analysis, we have to do an economic impact assessment on
the entire drinking water industry which is primarily small systems. At that
point, for the total national cost number, the small systems sometimes become a
large component. For instance, for ground-water treatment, small systems will be
the bulk of the cost (capital and operating), but I'm pretty sure for compliance
with ground-water treatment, for disinfection by-products, the large systems will
be the bulk of the cost, compared to the small systems.
Elizabeth Kawczynski - That's what I was thinking about. There was some comment
about trying to visualize Cincinnati treating the Ohio River water with membranes,
and we all kind of went, no I don't know if that will work, but yet, on the other
hand, there are smaller systems that would find it a very viable option, and it
would be cost-effective for them. That is what I am getting at in terms of Best
Available Technology—the cost.
Steve - We do analyze the cost for small systems, and we would look at membrane
technology. It seems to me to be feasible under certain conditions for small
systems, and by small I think we're talking at least, in this case up to 20 million
gallons per day all the way down to a few hundred people. I don't know what the
cutoff or absolute feasibility would be.
Dave Paulson - If I understood you right on the Best Available Technology for the
surface water treatment for microbes has not been promulgated, but you proposed
R0 as the Best Available Technology for inorganics?
Steve - For certain inorganics. We have two proposals out. One is called Phase
II that includes things like cadmium, chromium, and some of the traditional
contaminants such as nitrates.
Dave Paulson - Is there something published on that, that can be referenced if I
wanted to read the proposal?
Steve - A Federal Register Notice of Hay 1989 was called Phase II, which are the
more traditional inorganics in drinking water. Phase V was just proposed a few
weeks ago in the Federal Register.
Dave Paulson - If you've got these pending questions, such as radionuclides, does
that mean that there's more research required?
Steve - For those two phases, there's no research planned. In other words, we have
determined that if you use a typical reverse osmosis system (I don't know the
definition of it; I've been too far from it now), you can remove radium and uranium
in the radionuclide area. Actually, we haven't made that proposal.
Dave Paulson - You have information that it will work?
Steve - Yes. For radium there were studies ten or fifteen years ago on the West
Coast of Florida for radium.
Bill Conlon - Also, the Trace Metals Institute publication #604 and 607 address
radium removal by R0.
Steve - For the inorganic chemicals, reverse osmosis is pretty well established.
151
-------
Dave Paulson - I guess a real general question is where is the EPA on their plan
and their needs for the research for membranes? I had the feeling that a lot of
this meeting was to address research needs and what possible products should be
looked at here. I'm just trying to distill this down to what is going to happen
in the future. I should comment that I picked up in the lobby this project
summary, "A Study of Possible Economical Ways of Removing Radium from Drinking
Water", published April 1988, and it doesn't mention RO at all and that's why I
was a little confused when you said that.
Steve - That's because there were studies prior to that pretty well established.
I think that it's this whole issue of disinfection by-products and pathogenic
microorganisms and using membranes under a variety of conditions from something
very messy like the Ohio River down to something in Florida ground water.
Dave Paulson - Has your group in the Office of Drinking Water defined what specific
questions you need answers to and the date for the future research and the target
for the near term research to hit these dates, for instance, that the Oregon
citizens have pushed you to hit?
Steve - We have given Bob Clark, we give Bob Clark and his group every year a list
of research needs and among them is a need to have this kind of interaction:
number one, to bring in the industry and the people who are involved in design and
manufacture of these systems, and number two, to possibly look at existing systems
that are in place and how well they remove total organic carbon and other measures
of organics, possibly pathogenic microorganisms if that's feasible. And, to
possibly get out of this some ideas for future research. I think there is some
research continuing with removal of precursors for total trihalomethanes, but that
is the extent of our research. Dr. Taylor has been doing most of the research that
EPA is involved in. That's been the focus.
Dave Paulson - That's on the disinfection by-products, specifically. As a
potential supplier of this technology, I naturally have an interest in what areas
are not yet defined, but might be. For instance, what type of a membrane should
we be looking at that will be relevant to the EPA's current plans? In other words,
yes, we need something better for radium which is probably not really well rejected
as well as people would like it to be, that is the percentages are not high enough.
So is that something that has the status of yes, if we could have the rejection
we want then we could call membrane the BAT, or is membrane considered at all?
Is there some kind of written information that one could reference on the near-
term needs for hitting the regulations?
Steve - You mean as far as people complying with new regulations?
Dave Paulson - Well as far as in this case, specifically here, the BAT, you have
to decide some BATs and there are 83. Which ones are in the "probable", "no way",
and "need more information" categories? Is that organized?
Steve - We have a fact sheet for all the various phases of regulations that talks
about the chemicals and their Best Available Technologies. It starts with what
we consider Phase I, which is volatile organic chemicals, and carbon and aeration
are the BATs, and goes on to Phase II, with inorganics.
Dave Paulson - Are these publicly available, and are these appropriate for the
manufacturers?
152
-------
Steve - Yes. There is also an 800 number. We have a hotline you can call to get
any of these documents. It's called the Drinking Water Regulations Fact Sheet,
May 1990. Probably the biggest number of violations would be for nitrate, the
inorganic chemicals. It's really a trade-off between ion exchange which doesn't
work very well and reverse osmosis which doesn't work very well. If you were to
have some pesticides violations also, possibly it would be more economical to go
to reverse osmosis. That's kind of the market for treating ground waters that are
contaminated so to speak. A lot of the chemicals we regulate are fairly rare.
There are not that many violations of the standards. There are also other ways
of getting them out.
Bob Bergman - What is the most recent publication that EPA has involving the cost
of membrane systems? You say the cost is an important consideration, but I don't
think you have a cost report.
Steve - There is support documentation for both proposed regulations: the one for
May of 1989 and the one from July of this year -- Phase II inorganic chemicals and
Phase V inorganic chemicals. If you look at these Federal Register Notices, they
tell you how to obtain those documents. You can either obtain them by going to
the Regional Office that is closest to you or through the National Technical
Information Service (NTIS) or whatever it is. Basically we have cost models that
were developed under Bob Clark's auspices that allow us to make cost projections
for reverse osmosis and other technologies.
Bob Bergman - As I understand it, that cost model is limited to reverse osmosis,
and there isn't anything developed for ultrafiltration and microfiltration.
Steve - Yes, just for reverse osmosis. We don't have any cost models as far as
I know for ultrafiltration or any other of these technologies.
Nick Zelver - Is there an established procedure that EPA has in developing these
BATs, and if so, could you summarize the steps that are taken?
Steve - Sure. In order to determine what the Best Available Technology is we first
of all have to have a health goal. We call it the Maximum Contaminant Level Goal.
That's the level that we consider safe generally. For carcinogens, its zero, so
that makes it even tougher. For carcinogens, generally we either look at the
feasibility of measurement or the feasibility of treatment. So, in our case, we
are only going to talk about feasibility of treatment. We look at literature,
what's available in literature for any contaminant. Let's take nitrate. Is that
a good example? For nitrate, we find that certain kinds of ion exchange systems,
anion exchange systems with sodium chloride regeneration, will remove down to 10
milligrams per liter.
Nick Zelver - Is that an initial survey?
Steve - Yes. That's what the literature says. There's a plant out in California
that's doing this full-scale under some pilot studies, and maybe some other plants.
We then look at reverse osmosis, and I think there are some studies. Carol Ann
showed you, mentioned a study in Long Island where they removed more than 10
milligrams per liter of nitrate from the drinking water using reverse osmosis.
We look at that information, and we make a projection on what a model system would
look like in terms of design parameters, etc. We fit this into the mathematical
model that we have for costing, and we cost out the cost for 12 hypothetical system
sizes that range from 25 people up to over a million people. The Best Available
153
-------
Technology for setting the standard that would meet 10 comes from whether or
not it's roughly cost-effective for these larger size systems, say over a million
or over 50 thousand. It really doesn't matter because the curves are fairly flat
for most technologies up at that end. Generally, we are talking about additional
costs in the range of 2 or 3 dollars per thousand gallons of treated water, over
and above what they are already paying. So if they are already paying a dollar,
we are adding two or three dollars. We have been bouncing around these figures.
They are unofficial at this point, but this is affordability criteria. Once they
are official, at that point we will have selected what the Best Available
Technology is. It meets the health goal, people with larger size systems can
afford it, and it's been demonstrated at field scale for a specific contaminant.
Now for some contaminants, we don't have field-scale data, for instance beryllium.
Beryllium we hope behaves like some of the other metals. We run some pilot- or
even in this case possibly bench-scale studies here in Cincinnati. I see that we
have 92% removal, etc. So for beryllium, which doesn't occur very often by the
way, we say that reverse osmosis is also Best Available Technology, because the
behavior of beryllium is similar to say sodium where we have full-scale data.
Nick Zelver - You don't do that across the board. You look for field data where
you can get it.
Steve - Where we can get it, and we look at pilot- and bench-scale data to fit in
the odd ones. Particularly for the ones that don't occur very much. There's not
going to be field-scale data because, number one, no one was ever concerned about
beryllium until Congress required us to be. Congress has required us, EPA, by law
to regulate certain contaminants and beryllium is among them. It's sort of an odd
thing to be regulated, but apparently it has some toxicity, and it is in a few
drinking waters, whereas nitrate is long established as a drinking water
contaminant.
Vern Snoeyink - With respect to membrane systems for THMFP removal/precursor
removal, where does the Office of Drinking Water's evaluation of field performance
data stand? Do you feel comfortable that we have sufficient data to put membranes
in the Best Available Technology category?
Steve - We have very limited information actually. We had a consulting firm,
Malcolm Pirnie, Inc., from White Plains, New York and Parmus, New Jersey, collect
some information in what we call a cost and technology document evaluation. It's
really just on the information gathering end of this. They've found a limited
amount of information in the literature. Dr. Taylor's was one of them which EPA
sponsored, of course, Dr. Jacangelo's report, and there may be one or two others.
There is not a broad spectrum of reports. In fact, I saw information here that
I wasn't aware existed, for instance, simultaneously studying the removal of
Giardia and viruses and organic precursors. I think that is very critial informa-
tion to this technology evaluation, because we really are dealing with two things
happening at one time; removing precursors with membranes is one possibility. If
you are removing precursors, you are most likely removing viruses and Giardia.
which would then give you a much lower requirement for Ct values. So your
disinfection by-products consequently would be much lower and should be much lower.
That kind of information I think is lacking. It's a complex trail. I don't know
if you can follow all of my logic, but we are doing two things simultaneously,
actually three things: we are removing the bugs, we're removing the precursors,
and we are lowering the concentration/time values for disinfectants. Consequently,
we could get very low disinfection by-products. Or, as I suggested, our analysis
for setting the standard is focusing on conventional technologies of coagulation,
154
-------
filtration, possibly some form of carbon adsorption for organic precursors,
followed by chlorination to meet half a log of Giardia inactivation (it is the
chlorine, which is controlling), then some residual disinfection for the
distribution system in the surface water system. That would probably set the
number for total trihalomethanes somewhere in the range of 25 to 100, with 50 being
maybe a more likely number, but I really don't know at this point. This is
hypothetical. Given that information, I would think that getting some of these
studies published would show the pathogen removal, the precursor removal, and
possibly some formation potentials. Those three things would be very helpful for
us. Right now I think our literature survey has turned up maybe three or four
studies of ultrafiltration and these kinds of applications. I guess the other
question we have is application to surface water plants. Most of the studies we
have done have been on Florida ground water. I think there is one Florida surface
water study. Maybe a broader application would be useful including areas where
waters are colder, and this could possibly have an effect on performance, the
temperature fluctuations in the more temperate climates. It's a tough area for
us. We haven't dealt with you for all these years. We have been dealing with
granular activated carbon, ion exchange, reverse osmosis, and a few other
technologies, but this end of membrane technology, separation of organics and
larger molecules, is something fairly new for us. We've just begun to consider
it.
Larry Lien - Suppose I wanted to clean up the Ohio River, treating 200 million
gallons a day and I get 99% recovery. That means I've got 2 million gallons of
Bill's waste here. Is there any way that we can get some help evaluating what we
would do with that material? That's the crux of the problem. We talk about 99%
recovery in a membrane system as being marvelous, but on a system that size we
still have 2 million gallons a day to find a home for.
Steve - I think so. I think there are some technical questions, and there is a
new Federal law that allows Federal agencies to enter into a cooperative agreement
of some sort. It is called "Federal Technology Transfer Act". Bob is very
interested in getting into these kinds of arrangements, where we can work with
manufacturers or other people in industry on a joint problem, sharing costs
possibly, sharing benefits even, so if there is some new invention or patent, the
government can get some of the benefit from it. So that is something you should
talk to Bob Clark about. He is the head of our research program, whereas I am sort
of the regulatory guy. I write the regulations and try to assimilate all the
information and make some kind of decision that's based on a logical path, seeking
advice from a variety of experts. But the research really comes out of this
office.
Ben Lykins - Jim Goodrich and I have been working with some organizations trying
to get some research work going under that particular act. What we could do is
send the manufacturers that information and let them understand what the act is
about and see if they are interested in working with us. We could do that.
Bill Conlon - It would be nice in the case that was just cited that one could take
a simplistic approach of mass balance. Considering that you are going to give the
people a R0 treated water with superior quality now to drink, which will after the
public use ends up at the wastewater plant, you're going to mix in the concentrate
at the tail end of the wastewater plant, then you end up with the same water
quality as the raw water supply. How is that pragmatic anymore?
155
-------
Steve - The waste disposal issue is complicated by a whole bunch of factors.
Really, what our main interest is, is the performance of the technology. We are
hoping that a lot of these waste problems sort themselves out on an individual
basis, because they are all different. Different states are running things
differently, and from a Federal level, as long as we feel that you can somehow
dispose of the waste, then it's one of the technologies we need to consider when
we are developing these regulations for drinking water. We touch base with the
Office of Solid Waste, and they generally tell us this looks pretty good, and it
will probably pass the toxic extraction test.
156
-------
Jack Jorgenson - National Water Supply Improvement Association
I want to say that I am Executive Director of the National Water Supply
Improvement Association and I have had a long history with this technology and have
probably been around it longer than anyone in this room. I began in the late 50's.
I've seen a lot of ups and downs over the years and with a lot of down years over
the last ten. It is encouraging to see what's happened here, with a renewed
interest occurring, not only in this group but around the country. I think you
are seeing a growing proliferation of agencies, associations, committees and so
forth that are getting into this technology and that is good in one sense, but also
prescribes a need for getting together and thinking this out a little more as a
group to gather strength.
The National Water Supply Improvement Association has been around a long time,
it is the original association formed to foster this technology. That was 28 years
ago. Since then, we have been the supportive agency for the technology, the
research development, demonstration, technology transfer aspects of the technology.
The group is a combination of users of technology, (water district, state
agencies), the suppliers of equipment, (chemicals, membranes, pumps, valves,
pipes), and architect engineering firms that put things together and individuals
that are researchers or work for these various companies or finally just interested
in keeping up with new technology. The Association has its meetings every other
year. We are a national organization at this time, but have gone through a period
of being national, then international, and now we are back to a national. We are
currently affiliated with an international group. Within our group, we have an
affiliate within California that deals with primarily wastewater reuse. The last
international meeting was held in Kuwait in November of last year. NWSIA will
convene in Orlando, Florida for four days, and for the first time have a student
paper section, headed by Dr. Taylor. Another encouraging part of the business is
to see that these new young people start learning at the university level and not
having to learn it after they start working.
We have been involved in several technology transfer efforts over the years.
The Office of Technology Assessment report that came out a couple of years ago is
probably the first real generalized statement of the technology for a long time.
It was generated by Senator Simons from Illinois. The OTA called on our
organization along with others to help them put this together. After delivery to
the Congress, the Senator become more interested and drafted a bill that deals with
desalting/water reuse technology, setting up research and development grants,
demonstration projects. This bill has languished in the Congress for a couple of
years with not much interest primarily because Senator Simon did not participate
on the committees in water. But late last year in the appropriation hearings he
did insert in the appropriation bill a requirement that the Secretary of Interior
study and report back to the Congress on what the Interior Department would propose
for a research and development program in desalting and water reuse technology.
That report came out a couple months ago, its a blue report called "A Plan for
Improving Desalination in Water Treatment Technology". While it is not a wholesale
endorsement of a government financed research and development program at least it's
the first time in ten years that the administration has let a report of this kind
go through to the Congress. It's an indication that if the right type of program
is presented they might consider endorsing it. The Senator has picked that up and
is now continuing his interest but more importantly Senator Bradley from New Jersey
who heads one of the subcommittees on Environment and Energy, has become interested
and he has introduced a bill, a related bill, not too well written but being a
chairman of a key committee has now scheduled a hearing before the Congress and
157
-------
his committee for September. At that time the Simon interest and language of his
bill and the Bradley bill may be merged into a joint bill. According to Bradley's
people, the Senator doesn't get into these things without having some indication
that it is going somewhere. They feel that this bill has a chance this time to
become reality and that they are asking us for all the help that we can give them.
Another item that followed as a result of this interest from the Congress in
research and development is the Interior Department follow through on the Simon
proposal. The Department has asked NWSIA to come help them develop an agenda for
research and development needs, in the desalination water reuse technology. We've
entered into an arrangement with them and we will dedicate some of our professional
time as well as our member time that will culminate in a one day panel seminar to
be held one year from now, August of 1991 in Washington, DC. The International
Desalination Association will assemble at that time for their bi-annual meeting.
Dr. Taylor mentioned that small group of people both national and internationally
will be assembled at our meeting in Orlando to kick this thing off. Our Florida
meeting is in Orlando, Florida, 19-23rd of this month, we have about sixty papers.
Interest is worldwide, we'll have a group of about 30 or 35 people from the Middle
East, probably a total of 250 to 300 people. NWSIA is not big but we do
concentrate on a very small slice of the industry.
158
-------
MEMBRANE MANUFACTURERS PLANS
Ben Lykins - One of the questions that was asked, what is the research need, what
are the research needs of EPA, and I don't know, Steve tried to explain it, I'm
not sure whether we understand what they are at this time or not. Did he answer
your questions?
Dave Paulson - Yes he answered it in the sense that he gave me the references to
find, I guess. Some of the BATs like for radium have been determined and that's
good. It would be nice to see sort of a comprehensive list of them in this area,
that would be useful, but short of that I'm sure I can dig through the CFR
references.
Ben Lykins - Is there any possibility we could come up with a list of needs for
the manufacturers. Relative to membranes.
Steve Clark - Sure. Relative to membranes.
Dave Paulson - It would probably be both the need for what membranes could do to
accomplish something that you know right now, if the current ones can't, so that
would be product needs. And then just data characterization needs.
Ben Lykins - As we develop these needs you know we are in the research portion of
EPA and we only have a limited amount of funds so we can't do all the research
that's needed. So oftentimes the Office of Drinking Water has to rely on you
people as well as the researchers that presented this morning. So I think it's
important that Steve come up with those types of needs.
If we could just very quickly let the manufacturers speak a little bit, the
rest of us have had chance to talk. Even if it is generically, if you could tell
us what you anticipate on the horizon for your research relative to what Steve has
presented today and some of the others as well as are there new membranes coming
out, how effective are they, are they more cost effective, and so on such as that.
159
-------
MEMBRANE MANUFACTURERS
Terry Marsh - Dow Chemical
Dow's research, and when I say Dow I'm also incorporating Filmtec Corporation
research in membranes, is really centered into several areas, broad areas, one
would be applications research and one would be a product research. With respect
to the topics we've discussed today, we really are not involved in any application
research as far as disinfection by-products and trihalomethane precursors, not
directly. We have some supported programs that are going on for example Dr.
Taylor's work, and we will be happy to continue to do that, to supply membrane
elements to people that are interested in that type of applications approach. From
the product development standpoint, there are several things going on in developing
products that have been used to date, one is an ongoing development with our NF70
membranes products to improve the performance and consistency of this membrane even
though it's been around for four or five years now. It's still probably not well
characterized in term of the breath of its performance and in fact we suspect that
the performance indicators that we have on it in terms of basic performance as its
manufactured do not relate terribly well to the types of performance that are
going to be seen in the field under a wide range of conditions. We are trying to
get a better handle on that as a way of understanding predictability in the field.
There are some other membrane products that are similar, product line extensions
that may well have some applicability in this area. We are working on a membrane
product that would fall in-between the NF70 and the current BW30 operating probably
closer to 100 to 120 psi that would have higher overall rejections than NF70, but
still operate at pressures quite a bit less than 200 to 225. That may have some
benefit in rejecting materials that now tend to pass through the current nanofil-
tration products. It's not clear, we don't have any specific targets that we are
shooting for in terms of materials that reject better, that are not being rejected
well now, other than basic characterization with single-solute solutions. There
is also work going on in the other direction in terms of making more selective
membranes and whether they have any application barriers or not remains to be
seen. There is one product that will probably be coming out later this year,
introduced later this year that operates at about 130 psi and has 95+% rejection
of magnesium sulfate but only about 50% of sodium chloride. A looser membrane than
the NF70 type, molecular weight cut-off of something on the order of 300, that is
highly dependent on the actual chemical structure. My guess is that probably won't
serve the needs of this particular application but it may be something to take a
look at. There may be down the road another nanofiltration product that will fall
in between that and the NF70, which may have chlorine tolerance but I do not want
to make any promises at this point, others have thought early on that membranes
were more chlorine tolerant than they turned out to be so we do need to study that
one. I think one other area that I think was brought up today that we are looking
at in our central research group which is still at a very early stage and that is
trying to get a more fundamental understanding of membrane fouling and particularly
biological fouling, so we've got one person that is looking in that area trying
to get a better understanding of the mechanisms that are taking place. We have
no clear definition as yet as what that may lead to in terms of new products or
modifications of current products.
Mark Clark - Is your company willing to divulge things like molecular weight cut-
offs and even further, pore size distributions that you've characterized. Several
of the researchers here are interested in modeling and it is impossible unless
you know these things.
160
-------
Terry Marsh - We are certainly willing to divulge the data as we have it or
understand it. The molecular cut-off when you get into the nanofiltration,
ultrafiltration membranes is pretty dependent on how it is measured, to the extent
we have data on specific membranes that we are happy to share.
Mark Clark - Do you agree with Dave Paulson, that a lot of this information is in
the literature and it is just a matter of finding it?
Terry Marsh - I think that's true to an extent, I guess I'm not sure I can
specifically address the full range of membrane products and that no one can. I
think the extent of the range is on the RO and nanofiltration type membranes that
is available but then there are a lot more that may not be published, information
that's come in from specific users, or data that we have that is unpublished.
Dave Paulson - My comment to clarify my statement, I'd say that for any membrane
that's been commercialized for three years or more you can find the basic
chemistry, either in the patents or in papers that have been done. I feel
confident in making that statement. Certainly that for newer ones and ones that
are substantially different, the manufacturer may not want to reveal it. I know
they are, I have sat through paper presentations where other people have described
and compared the membranes, polymer chemistry, for the commonly used commercial
membranes compared to these. Hydronautics just published one in Ultrapure Water
Magazine where their different types of membrane chemistry are explained.
Hermann Pohland - Properties such as your porosity or molecular weight cut-off
point can vary in any one commercial product, if the product is sold on the basis
of salt rejection. It's just like a detergent manufacturer who doesn't always make
the same formulation. You could have changes in membrane structure that the
customer never hears about, and the commercial name may still be the same, since
salt rejection and flow remain the same.
Mark Clark - We are finding a pack of disk membranes from Amicon that have a
difference in flux between two of those disk membranes, by a factor of 2-6%.
Dave Paulson - The more void space and the larger the nominal pore size, the
greater the variation you will see.
Brent Cluff - Do you have your 11 inch, I heard you were going to come out with
11 inch elements, NF70's?
Terry Marsh - We have an 11 inch product for the BW30, it's one that has not been
used to a large extent, the DuPont data products are pretty much the predominant
ones that are used.
Brent Cluff - So you are saying that the NF70 will not be available?
Terry Marsh - I wouldn't say that. One I don't know what plans are being
considered as far as future product lines, we don't currently have a commercial
configuration for the NF70 in the 11 inch. But I can put you in touch with people
who are a little closer to that product development.
Jim Taylor - Is there any organization known to you that would possess
standardization literature published by manufacturers? Things like MWC, pore size
distribution, materials so that anyone could get more technical information on
membranes? This would provide a little better foundation for the researcher to
investigate various problems.
161
-------
Terry Marsh - I think the ASTM has started in that area, and I think you are more
familiar with that organization, particularly as it relates to the membranes, but
they have started to put together some standards that relate to membrane
application, so that's certainly one group, and that's the only one that comes to
mind at this point and time.
Hermann Pohland - You have to see this within the context of the business these
membranes are sold in. They are sold into desalination, for instance, where the
flow and the salt rejection are the main criteria, so that is what is being
measured, but nobody can afford to keep track of the possibly hundreds of other
membrane properties that you could measure.
Jim Taylor - Does anybody ever use an organic salt instead of sodium chloride and
magnesium sulfate to characterize molecular weight cutoff and rejection? I
wondered in terms of organics, considering both synthetic and natural organic
rejection, that is of timely interest in solving today's problems.
Dave Paulson - I can try to answer that because as far as I know the most likely
place that would occur is in ASTM and then that would be in the UF task group, and
I am the chair of that. I started that, I brought it back to life so to speak,
but believe me it's barely breathing. The ASTM is a slow moving organization and
I don't see it moving very fast to arrive at some kind of a standard test for
molecular weight cut-off. There is another group that is trying to advance that,
that will probably get to something sooner but that will be specific to the
application of biotechnology, because it's the ASTM Biotechnology Group, I think
E48 is the name. At any rate if you are interested in where they're at on it,
Subdash Sikdar at the National Bureau of Standards is the head of that group and
they've got a proposed standard for determining molecular weight cut-off on UF
membranes. Again, it is from their specific application perspective. Very low
pressure I would say, on UF.
Steve Ary - Even in UF, I think most of these measurements are approximations.
If I challenge a membrane with a molecule which I assume is globular, when I get
to the concentration polarization layer and through the membrane, how do I know
that it is not going to unfold into a long strand and pass nicely through the
pores? In other words, there are at least three pore measurement methods, but I
have not seen anything that would be meaningful. I think it is either an ad hoc
approach or we can go by the ASTM rule-book and commit the same error in the same
direction all the time.
Dave Paulson - The thing that I am sure about is that even if you test the same
membrane, the same piece of membrane, you can arrive at just about any percent
passage you want to get to if you run in the right mode, that is if you run them
at the right concentration, temperature, pressure and cross-flow velocity. You
can get all sorts of different answers, so you can come up with any numbers you
want if you know how to test. So the question will be what would be a good stan-
dard test, and a lot of people have their own ideas and I don't think there's a
lot of incentive for the manufacturers to come to an agreement on it. There
doesn't seem to be because it hasn't happened yet.
Jim Taylor - Or reporting test conditions.
162
-------
Dave Paulson - Exactly, and the test conditions are really so complex that a lot
of people don't really want to know all of them, and they haven't been well
reported. In fact it's just occurring now, it's in the process. I think if you
go to the NAMSs conference in August, you will probably see 20 or 30 papers and
poster sessions on that topic, and they will give a fairly good description of the
test methods. But somebody knowledgeable in that area could come in and see all
of the things that, in any given one, were done "wrong" or what variables weren't
defined, and how the results could be changed, because it's an incredibly complex
set of conditions that all can affect the results.
Hermann Pohland - There are many examples of flawed comparisons. Some of these
studies that have been done comparing spiral with hollow fiber modules overlook
the fact that hollow fiber modules usually run at 50% conversion while the spirals
will run at 10%, when they compare the results they make no valid comparison,
because if you ran the spiral at 50% it would give totally different results.
Something has to be done to avoid these comparisons.
Jim Taylor - If you are using them to develop the field, you would hope.
Bill Conlon - Another need for both consultants that are doing demonstrations and
researchers in comparing one membrane against another is as soon as a new product
is on the market from a manufacturer and as soon as they get enough field data
back, would be to know their recommendations for normalizing the membranes, use
it for the A and B values. A standardization formula for that particular membrane
that we could use, so we could be comparing apples with apples.
Hermann Pohland - But the ASTM method for data normalization is available.
Mark Clark - But does it work for the lower molecular weight cut-off?
Hermann Pohland - It doesn't work for organics but it is designed for inorganics,
something has to be developed for organics.
Mark Clark - Is there an ASTM method for surface charge particles?
Dave Paulson - Surface charge?
Mark Clark - Yes, that is something we are interested in also, because particulates
in certain cases will be charged.
Dave Paulson - No, to my knowledge there is none.
Mark Clark - Do you feel like in your field that techniques are being developed
for that?
Steve Ary - Streaming potential maybe.
Mark Clark - Well, plus or minus would tell us something.
Bob Bergman - The thing is on surface charge it changes too, it can change with
post treatments, a lot of things.
Mark Clark - What you start out with might not be what you end with.
163
-------
Dave Elyanow - Are you talking about membrane surface charge? Then that charge
is quite a variation. Depending on ionic strength and pH and you have to go to
the manufacturers.
Dave Paulson - The manufacturers haven't settled on one. One of the stumbling
blocks is that it will vary by application. On the other hand, maybe the use of
membranes for potable water treatment is going to be a large enough field that a
separate set of conditions for these types of tests can be run and can be deter-
mined for just that application. For instance, you want to know the molecular
weight cut-off, so you would approximate the typical concentrations and the typical
color bodies and design some tests for those, just under expected operating con-
ditions. Then you don't want just to test it on a piece of membrane under commer-
cially unrealistic conditions and then have to rely on that working for a commer-
cial element in commercial conditions-which are a lot less favorable than you are
going to get in a test cell in a lab, which will give you quite a bit different
guidelines. So it could be done, but it would have to be pretty detailed, and to
be effective it would be specific to potable water. Yet, why not do it? I think
that's a good area for the AWWA Membranes Processes committee to tackle, which
would be the test conditions that are closest to how the membrane is going to be
used, test conditions, and the separation requirements for that application.
Jim Taylor - Could you comment on the size of the testing apparatus you use for
smaller 2 inch diameter membranes? Can it be used instead of the larger 4 inch
diameter membranes for testing? How do you utilize results from large and small
scale membrane tests?
Dave Paulson - I think for totally dissolved solutes with low concentration
polarization effects, you can get a very good separation test with very small
"radial flow" setups. You do not get good flux predication, certainly not over
time, and it's hard to duplicate those conditions that you are going to run into
in the commercial elements. If you want to study either concentration polarization
or fouling patterns to predict flux over time, you have to go to a full size
element. Even there you have to be certain your conditions are as strenuous and
unfavorable as you will see in the system. The system, of course, varies from one
end to another in terms of fluid dynamics and pressure. So you may want to pick
the average and run it that way to predict, to scale up. It is pretty well
accepted as foolhardy to try scale up to large systems from small disks, especially
from the stirred cell systems.
Jim Taylor - What about doing velocity research on fouling control with small test
cells in terms of flux. Can we get a valid look at the effect of different
pretreatments on fouling with a small test cell?
Gerald Foremen - It just doesn't work.
Hermann Pohland - Evaluation of pretreatment techniques I don't think it works.
Gerald Foremen - I don't think it works if you are trying to evaluate pretreatment.
If you really want to do it as best as you possibly can, if you've got a large
system. Then you go in with four inch equivalent on a minimum array that's in that
system. Then you find out.
164
-------
Terry Marsh - I think you may have to clarify what you mean by the lab cell, and
what you are exposing it to in terms of feed water. If you are talking about a
small quantity of a feed water in a recirculating system you are not geting any
kind of dynamic performance as it relates to fouling. Unless there is a basic
incompatibility problem between the material in the water and the membrane. If
you were to take a small cell, and put it in the field and expose it to one certain
type of feed water then you may learn some things about the fouling, but you've
got to take the difference in geometry into consideration and they could be
considerable.
Jim Taylor - But in that type cell you are talking about a flow through continuous
flow cell.
Terry Marsh - Continuous flow cell, not a recycle basis, where you are putting
fresh in lieu of concentrate permeate.
Dave Paulson - I think a general method to use for new applications and waste
especially, is that you screen the commercially available membranes; maybe the
supplier does or the customer does. You screen them under one set of conditions
that sets you in the right direction. Then the next step is to do it in a small
size pilot system and try some different element configurations which vary in
design and find what works better. It is all relative. You don't want to build
in too much safety factor because of added cost but if the vendor has to be
accountable, he's going to demand testing, like Jerry said close to the actual
conditions. That's the way its done now, it's very imperical and practical. I
don't think anybody feels confident to do it another way.
Bill Conlon - I think that's apropos to performing jar testing at a coagulation
plant. It just leads you in the right direction. Pilot testing doesn't give you
as accurate data as using standard production devices in a demonstration program.
165
-------
Hermann W. Pohland - Permasep Products/DuPont
Permasep Products are the reverse osmosis desalination business of DuPont.
They are one among several membrane based businesses. These focus on chlor-alkali
production (Nafion membranes), liquid separations in food related applications
(orange and grape juice concentration), and gas separations. In the food and
beverage applications, DuPont has a joint venture with FMC Corporation and the gas
separations business is being pursued by a joint venture with the French Company
Air Liquide. Since I am in the marketing organization of Permasep Products, my
comments relate to the field of desalination primarily.
Permasep Products' research is driven by the goal to provide improved and more cost
effective desalination systems. Primarily, this is done by developing new mem-
branes with higher flow and rejection, but in addition we spend much effort to
improve the configuration, in which the RO membrane is being used. Permasep
Products offer the full spectrum of RO devices. Our seawater membrane is produced
in hollow fine fiber configuration (B-10) while our low pressure membranes for
brackish water are spiral elements (A-10).
Reverse Osmosis technology resembles in many ways ultrafiltration and
microfiltration, but since by now it is a mature business - Permasep Products
celebrated its 20th anniversary last year - RO does not require extensive pilot
testing anymore. RO has progressed to the point where you can come to us and I
guess to our competitors, too, and ask for a 10 MGD plant and in most cases, it
can be built without a pilot plant.
To return to the subject and to summarize: The focal point of our research
is development of more cost effective desalination systems using membrane techno-
logy. Specifically, Permasep Products are searching for desalination membranes
with better flow and better rejection, as well as, improved long-term operating
properties. Polyamide membranes pioneered by DuPont in RO have proven to be extra-
ordinarily durable, one of the goals of many researchers is to duplicate this dur-
ability and offer chemical inertness toward chlorine. However, to date, this goal
has not yet been achieved in spite of many efforts. Durability of the membrane
and selection of suitable polymers are very important, but the best membrane does
not always lead to the longest service life. Many other variables will influence
membrane life. Feed water pretreatment and operational skills can be of equal
importance to determine for how long an RO plant operates, before new membranes
are required. Permasep Products are proud to have plants in operation with B-9
membranes, which do not require replacement for more than ten years.
166
-------
Dave Paulson - Osmonics, Inc.
Osmonics is a minor player in the municipal potable water treatment area, out
of the membrane companies here. It's not a big part of our business, On the other
hand, a lot of what we do for engineering with membranes could be useful in this
area. Since we apply it in a more specialized and in a lot of cases more difficult
applications, some of the techniques apply and some of the approaches to solving
those problems should be valuable for making regular potable water treatment more
cost effective.
We do a lot of different things in terms of product development, and I am not
going to get into any of that. They are on the horizon; too speculative. But I
think that I would like to give my opinion on the area of product improvement and
development in general, and then of course you can assume Osmonics is interested
in those areas. I think that there are probably three areas that relate to potable
water and the first of them is just simply optimizing the engineering of the whole
system. For a long time, desalination was done pretty much one way, but now that
there are whole new sets of membranes, for instance, that can run at lower pres-
sure. You have to step back and look at the pumps. Are the flows, including those
for cleaning expectations, all really optimum or should they be revamped? I think
this is an area that's being done piecemeal and there are no big research project
that I'm aware of. It's being learned piecemeal and there is progress being made
to reduce the cost just based on how you design the system. Bigger breakthroughs
such as energy saving pumps, as well as the energy recovery on the pump, may have
been instituted for large desalination systems.
Hermann Pohland - It is a standard item now.
Dave Paulson - The second thing is improved element design, which Hermann talked
about too, will help reduce the cost. For instance, if you need very extensive
pretreatment the cost of the whole system goes up. Can you design a spiral wound
element, or hollow fiber or whatever, to be more self-cleaning or to have better
fluid dynamics for different applications? Are there more cost effective cleaning
methods such as backflow type, permeate back impulsing, etc. The optimum design,
would be so it simply doesn't foul.
That leads into the membrane treatment too, I think the third major area is
membrane development and will relate to surface chemistry. Charged membranes,
understanding the charges, putting added charges, I think that is the next big
step. The big step that has been taken is the thin-film composite where you have
such a thin active layer that the flux goes up, and you can get down to some very
small pores and get the kind of organic rejections that the composite membranes
now give. The next question is what can you do with the various chemistries,
especially in terms of controlling charges to improve the separation and make it
more economical. Take UF membrane and add charges to achieve salt rejection and
you make a softening membrane and still maintain a high flux at low pressure. When
that happens then the element is going to need a design change, as will the system
design. So those are three areas that I see on the horizon for the whole membrane
industry. They are likely to evolve in a lot of little steps.
167
-------
Gerald Foreman - UOP Fluid Systems
With respect to new membranes, Fluid Systems is always looking at new polymers
for our membrane systems. One of the areas we have become active in over the past
couple of years is in food separations. In this area we have done a significant
amount of work with ultrafiltration membranes. It is our intention to translate
some of this technology to the water side of the business. We are always looking
for new polymers.
The things that Dave Paulson of Osmonics said about new membranes, system
design and construction of the spiral are important. The use of low pressure
membranes in systems with high recovery rates preclude the use of classical RO
system design. The people who design and build the systems need to consider
interstage boosting pumps in order to compensate for the hydraulic pressure losses.
Otherwise, there isn't enough pressure remaining for the downstream elements.
There are a number of systems being built and installed today with this problem.
You might as well remove the downstream elements as they are contributing little
and in some instances, do as much damage as good.
In listening to the presentations today and developing a better understanding
of what is wanted, I'm not so sure that the approach of the RO industry to testing
is what you want or need. We have been interested in desalting the ocean and have
adopted the passage of sodium and chloride ions as our measure of success. For
your purposes, I believe that leakage flow (via a hole in the spiral or the
membrane itself ) is something you need to pay special attention to. Whether
others will admit it or not, there are 'holes' in the spiral and some of the salt
passage observed in a system is the result of these imperfections. This is the
major reason for the range of properties, particularly salt rejection, found in
spiral elements today. I can assure you that elements exhibiting below nominal
salt rejection have, in all probability, a hole in them somewhere. As membranes
come off the machine they are relatively free of imperfections. During the
translation from sheet membrane to the spiral some damage does occur and, depending
on fabrication techniques, a small number of minor holes may occur. There is not,
to my knowledge, a spiral supplier who will warrant zero leakage flow. In one of
the papers it was stated that they were surprised to see a certain species pass
through the membrane. This does not surprise me at all as the passage can probably
be explained by leakage flow. In my opinion, particularly with respect to
biological and/or radioactivity, the sodium or chloride ion may not be the best
measure of salt rejection as the membranes are not absolute barriers for these
species and are therefore poor measures of small leakage flow.
I also suggest that the EPA should contact, if it has not already done so,
the Bureau of Reclamation in Denver on its research efforts on the Yuma Project.
They have and are doing a lot of work with membranes, some of which is very
applicable to your requirements as I understand them. Mention of the 200 million
gallon per day Ohio River project brings to mind the 75 million gallon per day Yuma
Project. If you are in the area you should stop and visit the facility. Also,
the Denver Water Reuse Project and Orange County Water District's research work
should also be of interest to you.
168
-------
Steve Ary - Hydranautics
Just a few things. What I heard today was very interesting. It seems that
the main goal is to reduce trihalomethane precursors to a certain level. This is
in turn measured by the level of trihalomethanes generated by disinfecting the
purified stream. Trihalomethane precursors can have a wide range of molecular
weights. If I remove the high molecular weight material, the low molecular weight
material in the permeate can give rise to unacceptable levels of trihalomethanes.
So in other words, UF membranes alone cannot do the job and we may need membranes
with all cutoff ranges (UF-RO combinations for instance). Do we agree with the
FDA that an empirical (rule of thumb) 50 ppb level is acceptable? Is this level
acceptable to the EPA?
I can reduce precursors to the level that gives rise to less than 50 ppb
trihalomethanes for instance, Hydranautics plant at St. Lucie, FL is a 12 mgd plant
that reduces trihalomethanes but only by an RO membrane. Now does the EPA say that
there is a limiting value for trihalomethanes? Zero level would be ideal but you
may realize that it is very difficult to believe in that zero.
Based on this, I believe that one single UF membrane cannot do the job. If
only the high molecular weight stuff (and viruses) are to be removed, a UF membrane
can do it. I have to remove low molecular weight (300-500 Daltons) precursors as
well. Then, I really need an RO membrane. It is not even certain that it is a
very loose RO membrane. It depends on the feed. At St. Lucie, we use a rather
loose RO membrane, but there are no guarantees that we can repeat this somewhere
else.
And here comes a question, what kind of membranes do we visualize in these
applications? There are not very many possibilities. Organic chemistry or mother
nature, allows us to deal with a limited number of functional groups through which
we can crosslink these chemicals to create a barrier layer. The most readily
available functional groups were beaten to death by all of us. I think a new type
of membrane is needed.
I think it is a challenge to manufacturers and researchers to get away from
this merry-go-round. Finally, I would like to hear about guidelines from the EPA
what is really needed here. To reduce trihalomethanes, is it satisfactory to
achieve the 50 ppb level or 40 or 30 or whatever? If these levels are the measure
of my membrane that I do not have to worry about pore sizes or the chemistry of
the barrier layer. We could do the job first and we can go into details later.
169
-------
Dave Elyanow - Ionics, Inc.
We are primarily a supplier of electrodialysis equipment, but we also build
a lot of RO equipment and UF equipment. One thing that we do have a lot of
experience on in an industrial scale is surface water UF treatment at several
nuclear power plants where we're producing ultra pure water from surface water
sources using multiple processes, ultrafiltration, electrodialysis, and reverse
osmosis. When you look at the various processes that are available including
conventional treatment at some point you have to define what do you really need
to get a job done. Oust as you were saying, you need to know whether you need 50
ppb of THH in the effluent and are you going to do that by chloramination or are
you going to remove most of the organics. There have to be balances traded off
and this goes along with also the waste treatment aspect, because in many plants,
especially industrial there is a real problem with waste disposal. We have a
number of systems, electrodialysis systems, concentration RO effluents for waste
minimization. Then the question is what are you able to really dispose and how
are you able to dispose it. Also, we make equipment such as tubular microfiltra-
tion, that is combined with coagulation processes for printed circuit boards and
plating shop applications. Those are specialized waste but here you've got
specific heavy metal applications and so forth to deal with and you have the
question of waste disposal coming in as well. You take some of these plants and
you are going to be concentrating organics or even inorganic metals, I mean even
something like beryllium. What are you going to then do with the concentrate,
just dispose it in back with wastewater from the rest of the plant? This is
especially a question with radionuclides, I mean here you have a radioactive
discharge. Now this you know, you're not going to contain it for a thousand years
to allow it to dissipate, you have to make some decisions about what to do with
these wastes. One area we've gotten into at Ionics of course, is removal of
several of the inorganic parameters. We've demonstrated electrodialysis is a very
viable technology. There's over a hundred million gallons per day installed
worldwide capacity, and it's being used in a very practical manner. You have to
look at this as another alternate membrane technology along with ultrafiltration
and reverse osmosis. Further down the line we may get to volatile organics
pervaporation and other processes. I'm just going into some of the areas that we
are involved in. For organics removal you do get some organics removal with
electrodialysis as you do with ultrafiltration. It's not very high but there are
questions of looking at combination methodologies that might be appropriate for
given problems. Also, you have the question of some of the other parameters such
as nitrates and so forth that need regulation. We are working on having even more
selective membranes for nitrates. We have extremely durable membranes right now,
we have membranes that are working now that have been in existence for over 14
years and they are still operating in plants. We are always trying to improve the
durability. We have made electrodialysis membranes now so that they are reasonably
chlorine tolerant, take up to about 1/2 a part per million of chlorine, and we are
working on ways of making them even more chemically stable. We are very active
in electrodialysis and in using multiple membrane processes and then building
systems.
170
-------
Nick Zelver - Ionics, Inc.
I want to make a comment related to the Best Available Technology designa-
tions. To my understanding this really doesn't mandate a particular technology
but does become very critical in what people will look at, so it is very important
in the future what people use. What our concern is when we don't have definitive
field data, and the question of beryllium came up, but also particularly with other
inorganics, with the range or organics, and the often site specific nature, what
type of membrane to use and what type of pretreatment and such. That incorporation
of doing pilot studies is built into the Best Available Technology, that we just
don't rule out a technology because we don't have field data. I think that needs
to be a little looser definition, or opened up, be a little broader, that in some
cases yes, it's a great technology, but you should do pilot studies and determine
a little closer.
171
-------
Jim Taylor - We have received the abstracts for this Membrane Conference that is
going to occur in Orlando in March, and there weren't hardly any manufacturers that
submitted. It would be nice if we could get some abstracts from the manufacturers
to look at and work into the program, dealing with subjects that we've talked about
here today. Certainly productivity, studies that you've done, types of materials
that you are looking at, research needs as you see them, possibly applications for
the technology development program, I believe you spoke of Ben. So I would like
to appeal to the manufacturers that are here and other manufacturers that aren't
here that you may know to try to solicit some submittals and get them to John
Brittan at AWWA in Denver as quickly as possible so that we could continue to have
these forums for interchange between EPA and researchers and people in the water
industry, manufacturers and consultants. I think that's helpful and very good.
172
-------
ACKNOWLEDGEMENT
The author thanks Sandra Dryer for recording and transcribing this workshop
and Toya Jackson for helping type the proceedings.
This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication. Mention of trade names or commercial products does
not constitute endorsement or recommendation for use by the USEPA.
173
------- |