EPA-670/2-73-100
December 1973
Environmental Protection Technology Series
Applications of Reverse Osmosis
To Acid Mine Drainage Treatment
National Environmental Research Center
Office of Research and Development
US. Environmental Protection Agency
Cincinnati. Ohio 45268
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and
Monitoring, Environmental Protection Agency, have
been grouped into five series. These five broad
categories were established to facilitate further
development and application of environmental
technology. Elimination of traditional grouping
was consciously planned to foster technology
transfer and a maximum interface in related
fields. The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
1. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL
PROTECTION TECHNOLOGY series. This series
describes research performed to develop and
demonstrate instrumentation, equipment and
methodology to repair or prevent environmental
degradation from point and non-point sources of
pollution. This work provides the new or improved
technology required for the control and treatment
of pollution sources to meet environmental quality
standards.
-------
EPA-670/2-73-100
December 1973
APPLICATIONS OF REVERSE OSMOSIS TO
ACID MINE DRAINAGE TREATMENT
By
Roger C. Wilmoth
Crown Mine Drainage Control Field Site
Box 555
Rivesville, West Virginia 26588
Program Element IBBO^O
NATIONAL ENVIRONMENTAL RESEARCH CENTER
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
CINCINNATI, OHIO 1*5268
For sale by the Superintendent of Documents, U.S. QoYemment Printing Office, Washington, D.C. 20402 - Price $2
-------
EPA REV/IEU NOTICE
This report has been reviewed by the
National Environmental Research Center,
Cincinnati, and approved for publication.
Mention of trade names or commercial prod-
ucts does not constitute endorsement or
recommendation for use.
ii
-------
FOREWORD
Man and his environment must be protected from the adverse
effects of pesticides, radiation, noise and other forms of pol-
lution, and the unwise management of solid waste. Efforts to
protect the environment require a focus that recognizes the inter-
play betueen the components of our physical environment—air, water,
and land. The National Environmental Research Centers provide this
multidisciplinary focus through programs engaged in
• studies on the effects of environmental
contaminants on man and the biosphere, and
• a search for ways to prevent contamination
and to recycle valuable resources.
The studies in this report investigate techniques for pollution
abatement from the acid waste of the coal mining industry and, at
the same time, provide a means for recovery of a high-quality water
which would be suitable for domestic or industrial use.
A. U. Breidenbach, Ph.D.
Director
National Environmental
Research Center, Cincinnati
111
-------
ABSTRACT
Spiral-wound reverse osmosis systems were tested on acid mine
drainage discharges at four different locations (Norton, West
Virginia; Morgantown, LJpst Virginia; Ebensburg, Pennsylvania; and
Mocanaqua, Pennsylvania) uhose water quality characteristics were
quite varied. In addition, comparison studies were made of the
hollow-fiber and tubular systems at Mocanaqua and of the spiral
and hollow-fiber systems at Norton.
At all sites, the limiting factor in high recovery operation
was calcium sulfate insolubility. Generally, calcium sulfate fouling
occurred when vPmc/(2.16 x ID +) exceeded 2.0, where Pmc = product
of the molar concentrations of calcium and sulfate in the brine stream.
In all tests, product water was of near potable quality. Neu-
tralization was required in all cases to elevate pH and, in some
cases, to remove residual iron and manganese.
A "neutrolosis" process was developed in the course of these
investigations and constituted a major technological advance in
reverse osmosis treatment nf acid mine drainage. The nputrolosis
process, a combination of reverse osmosis and neutralisation,
achieved water recoveries near 99 percent at the Morton site while
producing a high duality product.
At Mocanaqua, the hollow-fiber and spiral systems exhibited com-
parable performance. Tubular performance, though adequate, was not
quite equal to the other systems.
At Norton, the spiral-wound system performed normally but the
hollow-fiber system experienced major colloidal and iron fouling
problems which were never completely overcome.
Operation of both hollow-fiber and spiral-wound systems at
Mocamqua under optimized flow conditions at 75 percent recovery
iv
-------
yielded log-log flux decline slopes near -D.D1 for 2,ODD hours of
operation.
Pretreatment at all sites consisted of ten micron filtration.
Ultraviolet disinfection, acid injection, or both, were necessary
at some sites to inhibit iron oxidation and precipitation.
This report was submitted in fulfillment of Environmental
Protection Agency Project 1^010 TMC. Work uas completed as of
June 1972.
-------
CONTENTS
Page
Foreword iii
Abstract. iv
List of Figures. viii
List of Tables x
Acknowledgements xiii
Sections
I Conclusions 1
II Recommendations . 3
III Introduction 5
IV Procedures 8
Test Equipment 8
Analytical Procedures 12
Calculations 12
V Results 18
Maximum Recovery Studies, Reverse Osmosis and
Neutrolosis 18
Two-Stage Reverse Osmosis System *+0
Morgantoun i*K Ferrous Iron Spiral-Wound Study...... ^5
Ebensburg ^H Ferrous Iron Spiral-Wound Study 53
Norton Ferric Iron Long-Term Spiral-Wound Study.... 71
Mocanaqua Ferrous Iron Studies Comparing Spiral-
Wound, Hollow-Fiber, and Tubular Units 82
Norton Ferric Iron Studies Comparing Hollow-Fiber
and Spiral-Wound Units 118
VI Discussion 138
Calcium Sulfate Precipitation - Maximum
Recovery Prediction 138
Comparison of Spiral-Wound, Hollow-Fiber, and
Tubular Systems 1^2
Significance of Rejections 150
VII References 15**
VIII List of Inventions and Publications 156
IX Glossary 157
VII
-------
LIST OF FIGURES
No..
1 Spiral-uound module configuration ........... ............. 9
2 Flow diagram for 37850 I/day (10,000 gpd) spiral-
uound reverse osmosis unit ............................. 1^
3 Hollow-fiber reverse osmosis module ...................... 13
4 Effect of recovery on brine concentrations ............... 15
5 Effect of log-log slopes on flux (all fluxes were
100% @ 1 hour) ......................................... 1?
6 Membrane performance during RO test number One at
91 percent recovery .............................. • ..... 23
7 Floui diagram for neutrolosis tests ....................... 29
8 Membrane performance during neutrolosis test (number
four) operating at 91 percent unit recovery and
98.8 percent system recovery ........................... 33
9 Membrane performance during neutrolosis test (number
five) operating at 78 percent unit recovery and
98.7 percent system recovery ................ . .......... 37
10 Two-stage RO treatment with intermediate brine
neutralization .................................... ..... 1+2
11 Flow diagram for 4K spiral-wound reverse osmosis unit.... *+6
12 ( A P) Pressure drop across tubes during test number
one ........ . .................... . ...................... 56
13 Membrane performance during Ebensburg test at 84
percent recovery ..... . ................................. 57
14 Membrane performance and AP history for Ebensburg
RO test number four at 50 percent recovery ............. 66
15 Flux trends for Norton 3000 hour spiral-wound RO
study at 13 percent recovery ........................... 72
16 Membrane performance during 4K reverse osmosis SOOO^-hr
study at (Morton ........................................ 76
17 Spiral-wound reverse osmosis system arrangement
at Mocanaqua, Pa ....................................... 87
18 Hollow-fiber reverse osmosis system arrangement
at Mocanaqua, Pa ....................................... 88
19 Tubular reverse osmosis module configuration at
Mocanaqua, Pa .......................................... 90
viii
-------
No.
20 Phase I module arrangement for tubular reverse
osmosis study at Mocanaqua, Pa 91
21 Osmotic pressure-conductivity relationships at
Mocanaqua, Pa 92
22 Phase I spiral-wound reverse osmosis unit flux
rates at Mocanaqua, Pa 95
23 A summary of reverse osmosis flux trends at
Mocanaqua, Pa 96
24 Operational history of hollow-fiber permeator
#691 during the Mocanaqua study 104
25 Tubular and hollow-fiber reverse osmosis unit flux
trends during the Mocanaqua study 105
26 Flux and AP history for the hollow-fiber phase I
study at Norton 120
27 Flux trends during the two permeator hollow-fiber
phase I study at (Morton 121
28 Flux history for the two first-stage hollow-fiber
permeators 125
29 Flux trends for 6K hollow-fiber phase II study @
69.7 percent recovery 126
30 Osmotic pressure-conductivity relationship of
Grassy Run at Norton..... 132
31 Total 4K spiral unit operating history 133
32 Maximum operating conditions to obtain potable
product for iron limits of 0.30 mg/1 151
33 Maximum operating conditions to obtain potable
product for manganese limits of 0.05 mg/1 152
IX
-------
LIST OF TABLES
Mo. Pa3.e
1 Typical Grassy Run Water Quality - 1969 .................. 19
2 Operating Parameters for 10K (Morton RO Study at
91 Percent Recovery, Test Number One ........ . .......... 21
3 Chemistry Analyses for Test Number One ................... 22
U Operating Parameters for Split Recovery Reverse
Osmosis Test (Test Number Tuo) ......................... 25
5 Chemistry Analyses - Split Recovery 10h Reverse Osmosis.. 26
6 Operating Parameters for Neutrolosis Study at 91
Percent Reverse Osmosis Recovery (Test Number Four).... 30
7 Chemistry Analyses for Neutrolosis Test at 91 Percent
RO Recovery (Test Number Four). ......... ....... ........ 31
8 Operating Parameters for Neutrolosis Study at 78
Percent RO Recovery (Test Number Five) ................. 35
9 Chemistry Analyses for Neutrolosis Test at 78
Percent RO Recovery (Test Number Five) ................. 36
10 Blended Feed, Brine and Product Chemistry Analyses
for 10K Studies. ....................................... 39
11 Operating Parameters for 50 Percent Recovery ^K RO
Study on Neutralized Brine ............................. U3
12 Chemistry Analyses for 50% Recovery <+K RO Study on
Neutralized 1DK Brine .................... .
13 ^+K Reverse Osmosis Operational History for Morgantoun
Ferrous Iron Study at the Arkuright Mine
Ik Chemical Analyses for ^K Reverse Osmosis Study at
Arkuright Ferrous Iron Site (Unit Recovery 50 Percent). UQ
15 Flux and AP Values for Arkuright Ferrous Iron Study ..... 50
16 Chemical Analyses of Arkuright Brine Precipitate ......... 52
17 Chemical Analyses of Solids in Arkuright Study
Unit after Shutdown .................................... 52
18 Blended Feed and Brine pH's at Ebensburg ................. 58
19 Operating Parameters for Test Number One at
Ebensburg, Pennsylvania ........ .... .............. • ..... 59
20 Chemistry Analyses for Test Number One Where Average
Recovery = 83.6% ....................................... 60
-------
g^ age
21 Flux and AP Values for Ebensburg Test Number Two 62
22 Operating Parameters for Test Number Tuio at
Ebensburg, Pennsylvania * 63
23 Chemistry Analyses for Test Number Four Where Average
Recovery = 84.0% 64
2k Operating Parameters for Test Number Four at
Ebensburg, Pennsylvania 68
25 Chemistry Analyses for Test Number Four Where Average
Recovery = 53.2% 69
26 Comparison of Mocanaqua and Ebensburg RO Test Results.... 70
27 Operating Parameters for 3000 Hour 4K Norton Study
at 73 Percent Recovery 73
28 Norton 4K 3000 Hour Chemical Analyses 74
29 Chemistry Analyses of Costing Material on 4K
Reverse Osmosis Membrane , 79
30 Chemistry Analyses for RO Product Neutralization 81
31 Typical Raw Water Quality Characteristics of
Mocanaqua Discharge 83
32 Ferrous Iron Oxidation Control Study, Mocanaqua, Pa...... 85
33 Operating Parameters for Spiral-Wound Phase I Study
at 75 Percent Recovery at Mocanaqua, Pennsylvania. 94
34 Operating Parameters for Spiral-Wound Phase I Study
at 84 Percent Recovery at Mocanaqua, Pa 98
35 Chemical Analyses for Spiral-Wound Studies at
Mocanaqua, Pennsylvania 99
36 Operating Parameters for Spiral-Wound Phase II Study
at 75 Percent Recovery at Mocanaqua, Pennsylvania 100
37 Operating Parameters for Hollow-Fiber Phase I Study
at 75 Percent Recovery at Mocanaqua, Pennsylvania 103
38 Operating Parameters for Hollow-Fiber Phase II Study
at 75 Percent Recovery at Mocanaqua, Pennsylvania 108
39 Operating Parameters for Hollow-Fiber Phase II Study
at 85 Percent Recovery at Mocanaqua, Pennsylvania 109
40 Chemistry Analyses for Hollow-Fiber Studies at
Mocanaqua, Pennsylvania 110
41 Chemistry Analyses for Tubular Studies at Mocanaqua,
Pennsylvania 113
XI
-------
Mo.
42 Operating Parameters for Tubular Phase II Study
at 43 Percent Recovery at Mocanaqua, Pennsylvania 114
43 Salt Passage for Tubular Studies at Mocanaqua, Pa 116
44 Calcium Sulfate Molar Solubility Products 118
45 Operating Parameters for Two Permeator Hollou-Fiber
Study at 72 Percent Recovery 122
46 Operating Parameters for Norton Hollouj-Fiber 3-
Permeator Phase II Study at 70 Percent Recovery 127
47 Chemistry Analyses for Norton Study (1972) 129
48 Operating Parameters for Norton Spiral-Wound 4K
Study at 69.6 Percent Recovery 135
49 Effect of Neutralization on RO Product Quality 137
50 Reverse Osmosis Recovery Limitations Due to Calcium
Sulfate Fouling 140
51 Summary of Raw Feed Chemistry Analyses 141
52 Comparison of Predicted Maximum Recovery with
Empirical Estimates of Maximum Recovery 143
53 Comparison of Water Production Capabilities Observed
During Mocanaqua Studies 144
54 Relative Cost Comparisons from Mocanaqua Study 146
55 Comparison of Membrane Performance at Mocanaqua 148
56 Projected Membrane Performance - Mocanaqua Studies....... 149
xii
-------
ACKNOWLEDGEMENTS
The author wishes to thank the following people for their help-
ful contributions and willing assistance during the course of these
studies:
Ronald D. Hill, Chief, and Eugene F. Harris, Deputy Chief, EPA
Mine Drainage Pollution Control Activities, Cincinnati, Ohio;
Robert B. Scott, Chief, James L. Kennedy, Chemist, Roger A. Dean,
Maxine K. Cooper, Curtis L. Corley, and J. Denver Tingler, EPA
(Morton Mine Drainage Field Site, Norton, West Virginia; Loretta J.
Davis, EPA Croun Mine Drainage Control Field Site, Rivesville, W. Va.;
James H. Sleigh and Seymour Kremen, Gulf Environmental Systems, San
Diego, California; Victor J. Tomsic and David Moyes, E. I. DuPont de
Nemours, Glascow, Delaware; Donald G. Mason, formerly of Rex Chainbelt,
Milwaukee, Wisconsin; Frank 5. Dilliplane, West End Coal Pockets,
Mocanaqua, Pennsylvania; Wade Elliott, Bethlehem Mines Corporation,
Ebensburg, Pennsylvania; Vincent Ream, Christopher Coal Company,
Osage, West Virginia; and Charles T. Holland, retired Dean, West
Virginia University School of Mines, Morgantown, West Virginia.
Special thanks are extended to J. Randolph Lipscomb and Alvin W.
Irons, EPA Norton Mine Drainage Field Site, Norton, West Virginia.
xiii
-------
SECTION I
CONCLUSIONS
The following conclusions summarize the results observed from
tests of spiral-wound, hollow-fiber, and tubular reverse osmosis
systems made on acid mine drainage discharges at four sites (Norton,
UJ. Va., Morgantown, kl. Ua.f Ebensburg, Pennsylvania, and Mocanaqua,
Pennsylvania):
1. Reverse osmosis has been shown to be an effective method of
treating acid mine drainage with the production of high quality water.
2. Product water from all sites would require neutralization to
increase pH and, in some cases, to remove residual iron before pota-
ble standards could be met* At sites where manganese concentrations
exceeded 10 mg/1, product quality would require neutralization to
pH 9-10 to effect manganese removal to meet U. S. Public Health
Service potable standards. Reacidification would then be required
to reduce pH to acceptable limits.
3. Recovery was limited at each site by calcium sulfate insol-
ubility. In general, calcium sulfate fouling occurred when
VPmc/(2.16 x 10~ ) was greater than 2.0 where Pmc = product of molar
concentrations of calcium and sulfate in the brine stream.
4. All test sites considered, performance of the spiral-wound
system was superior to that of the hollow-fiber and tubular units.
Although the hollow-fiber performed on a par with spiral-wound at
Mocanaqua, the hollow-fiber system had considerable difficulty in
treating the Norton ferric iron surface water acid mine drainage
because of iron and colloidal fouling problems.
5. Neutrolosis (the blending of neutralized brine supernatant
from the reverse osmosis unit back into the feed to the unit) has
been shown to be a promising process for ubtaining maximum recov-
eries and solving the brine disposal problem.
1
-------
6. Unless the neutrolosis concept were used, treatment of the
brine before discharge would be required in all cases of reverse
osmosis treatment of acid mine drainage.
7. Operation of the spiral system at i+00 psi with brine/pro-
duct (b/p) flou ratios in excess of 10:1 per module provided sig-
nificant improvements in flux stability as compared to higher
pressure, louer b/p ratio operation.
-------
SECTION II
RECOMMENDATIONS
It is recommended that additional research be implemented in
the following areas:
1. Determination of membrane life under acid mine drainage
conditions and development of reliable cost figures.
2. Development of a calcium sulfate precipitation inhibitor
uihich mould allow significant increases in attainable reverse
osmosis recovery levels.
3. Development of techniques for fouling removal (i.e., removal
of iron and calcium sulfate precipitation and organic deposition).
k. Optimization of neutrolosis process and testing under a
variety of conditions to determine scope of application.
5. Evaluation of reverse osmosis for removal of heavy metals
commonly found in "hard rock" type mine drainage.
It is also recommended that future system comparison evaluation
work not correct flux for pressure losses from internal piping and
intrinsic brine flow resistance since such values are characteristics
of the system involved and deleting these considerations would only
tend to cover up losses in process efficiency.
-------
SECTION III
INTRODUCTION
Betueen 1968 and 1972, the staff of the U. S. Environmental
Protection Agency's (EPA) Morton Mine Drainage Treatment Field
Site investigated reverse osmosis (RO)* treatment of acid mine
drainage (AMD).
Beginning in 1968, EPA (then Federal Water Pollution Control
Administration - FUPCA) cooperated with the Office of Saline Water
(OSLJ) in research at Norton, West Virginia to demonstrate the
feasibility of reverse osmosis treatment of a ferric iron acid
1 2
mine drainage water. ' During these initial studies, OSW supplied
the spiral-wound RO equipment and a consultant. EPA provided ana-
lytical support and assisted in operating the unit. Most of these
studies uere conducted at water recovery rates between 50 and 75
percent, although one test was as high as 80 percent, because of
the fear of fouling the membranes with iron and/or calcium sulfate
precipitation. At recoveries greater than 70 percent, calcium
sulfate exceeded the theoretical saturation concentration and a
potential danger of fouling the membranes was present.
It was concluded from these studies that water recoveries up to
80 percent could be achieved on the Norton ferric water with no un-
controllable iron or CaSO. fouling. Salt rejections between 97 and
99 percent were obtained.
The Environmental Protection Agency was deeply concerned with
the waste brine which constituted from 20 to 25 percent of the wa-
ter treated by an RO unit operating at 75-80 percent recovery. One
approach to decreasing this problem would be to increase the recovery
rate to a maximum, thus producing a minimum volume of brine. Towards
*See Glossary for definition of terms.
5
-------
this goal, EPA entered into a contract with Gulf Environmental Sys-
tems Company in 1969 to continue tests at Norton with emphasis on
high recovery operation. DSld supplied a portion of the RO equip-
ment. EPA conducted the tests with consultation of Gulf Environ-
mental Systems and portions of the results are discussed in this
report. Recoveries above 90 percent were obtained during these
studies in short duration runs.
Concurrent with the Gulf contract, EPA, through a grant to the
Commonwealth of Pennsylvania, contracted with Rex Chainbelt, Inc.,
to study the feasibility of reverse osmosis treatment of a ferrous
AMD discharge at Mocanaqua, Pennsylvania, using a tubular RO unit.
Although the water quality characteristics of the Mocanaqua dis-
charge were similar to the Norton water except for the ionic state
of the iron, severe fouling was observed at moderate recovery levels
/4
at Mocanaqua. The implication therefore existed that ferrous iron
was, for some unknown reason, more difficult to treat than a compar-
able ferric discharge. Although Riedinger and Shultz in 196S had
studied spiral-wound type RO treatment of a ferrous discharge near
Kittanning, Pennsylvania, and reported no severe fouling, it was
desirable that ferrous situations be more thoroughly investigated.
For this purpose, the scope of work of the on-going Gulf Environ-
mental Systems contract was modified by mutual agreement to include
studies on ferrous waters in addition to the (Morton ferric site.
The first ferrous site studied by EPA was a severely polluted
AMD discharge near Morgantown, West Virginia. Even at 50 percent
recovery, severe fouling occurred at this site. Due to the extreme
concentrations at Morgantoun, the situation was not comparable to
that at Mocanaqua. A site with conditions comparable to Mocanaqua
was located near Ebensburg, Pennsylvania, and the spiral-wound unit
was tested there in 1970. (\lo fouling was observed at 50 percent
recovery and only minor fouling occurred at 8^ percent recovery.
-------
Follouing the Ebensburg test, the spiral unit uas returned to
Norton for a long-term, 75 percent recovery study which lasted
3,012 hours and in which some major fouling occurred.
The (Morton Mine Drainage Field Site also had an active research
program concerned with neutralization of AMD. The staff conceived
that the reverse osmosis process could be combined uith the neutral-
ization process to produce a maximum amount of high quality uater
uith a minimum volume of waste. The process conceived uias operation
of the reverse osmosis unit at maximum recovery (90 percent at Norton),
neutralization of the brine, and recycle of the neutralized brine
supernatant uater back into the feed to the reverse osmosis unit.
Thus, this system uould produce only product uater plus a small
amount of neutralized sludge containing the iron, calcium, sulfate,
aluminum, etc. This process uas named neutrolosis (combination of
reverse osmosis and neutralization) and uas tested by EPA at Norton
during 1970.
In 1971, EPA, again through a grant to the Commonuealth of
Pennsylvania, contracted uith Rex Chainbelt to further evaluate the
fouling observed in the 1969 Mocanaqua studies. Rex Chainbelt re-
turned to Mocanaqua uith an improved tubular unit and a hollou-fiber
unit. EPA installed a spiral-uound unit to operate alongside the
other tuo under identical conditions. Severe fouling occurred only
on the tubular system equipped uith a membrane uith a salt rejection
of 96-98 percent. Both the spiral-uound and hollou-fiber units per-
formed extremely uell at 75 percent recovery. Very small flux losses
uere observed for the spiral and hollou-fiber units.
After the 1971 Mocanaqua studies, both the spiral-uound and
hollou-fiber units uere returned to EPA's Norton facility for final
tests.
Results of the EPA studies are presented in this report.
-------
SECTION IV
PROCEDURES
TEST EQUIPMENT
TUJO spiral-uound type reverse osmosis units uere used in the
course of these studies. Both uere manufactured by Gulf Environ-
mental Systems. The initial unit (owned by OSld) uas a 37,850 liter
per day (10 K - 10,000 gallons per day nominal product flou) unit
consisting of five ^ in x 10 ft (1.58 cm x 3.05 m) Schedule UO steel
pipe pressure vessels which had been lined with an epoxy to prevent
corrosion. A Moyno high pressure progressive cavity stainless
steel pump mas used to supply water to the unit at pressures up to
o
5,517 kiloneutons per square meter - kN/m (800 psig).
Each of the five pressure vessels contained three spiral-uaund mod-
ules (Figure 1). Each module uas 0.92 meters long (3 ft), 9.53 centi-
2 2
meters (3.75 in) in diameter, and contained b.65 m (50 ft ) of
modified cellulose acetate membrane. The three modules were placed
in series in each pressure vessel (tube). There uere tuo effluents
from each module (or tube of modules), i.e., brine and product uater.
Typically, the brine from one module (or tube of modules) served as
the feed to the next module (or tube of modules) and uas progres-
sively deuatered in this manner- Each vessel therefore contained
2 2
13.96 m (150 ft ) of membrane for a five vessel unit total of
69.8 m2 (750 ft2).
The flou diagram for the 10 H unit is presented in Figure 2.
Pressurized sand filters and 10 micron cartridge filters uere used
to remove suspended solids from the feed uater before the uater
entered the RO unit. Filtration uas necessary -as suspended solids
can clog the narrou brine channels and foul the RO membranes.
For these studies, the pressure vessels uere arranged in a
2-2-1 array in order to maintain, as nearly as possible, a uniform
8
-------
Purified
Water
Outlet
Membrane Module
Concentrate Outlet
Seal
Pressure Vessel
Feed Side Spacer
Roll to
Assem ble
Feed Flow
Permeate Ou
Brine Out
Permeate Side Backing
Material with Membrane on
Each Side and Glued Around
Edges and to Center Tube
Figu re 1
Spiral-wound module configuration
-------
10 MICRON
CARTRIDGE FILTERS HIGH
PRESSURE
PUMP
TUBE
TUBE 2
TUBE 3
TUBE 4
i
PRODUCT
DISCHARGE
TUBE 5
RECYCLE BRINE
GRASSY RUN
BRINE DISCHARGE
Figure 2. Flow diagram for 37850 I/day (10,000 gpd)
spiral -wound reverse osmosis unit.
-------
brine flou through each vessel. The first tuio vessels were in
parallel. Brines from these vessels uere combined to serve as the
feed to vessels 3 and 4, uhich uere in parallel. Brines from ves-
sels 3 and 4 uere then combined to serve as feed to vessel 5.
In order to increase recovery beyond 70 percent and yet main-
tain sufficient brine flou velocity and turbulence in each module
to prevent 'boundary layer1 precipitation, it uas necessary to re-
cycle a portion of the brine into the feed to the RO unit. In this
case, the spiral unit membranes uere actually treating a blended
feed (mixture of raw water and recycled brine) rather than the rau
water.
Inlet pressure and pressure drops across each tube uere measured
on pressure and differential pressure gages. Internal flou measure-
ments uere made using venturies and differential-pressure flou gages.
External flous uere measured volumetrically. As product flou uas
strongly temperature dependent, careful temperature measurements
uere made using a laboratory thermometer. The spiral 10 K unit uas
not mobile and therefore uas used solely at Norton.
The second spiral-uound unit produced 15,140 liters per day
(4,000 gpd) of product uater (4 K unit) and consisted of three
4 in x 10 ft (1.58 cm x 3.05 m) pressure vessels arranged in a 2-1
array. Operation and construction of this unit uas the same as the
10 K unit previously described. This 4 K unit uas ouned by EPA and
due to its size, uas especially suited to portable field testing.
Several feed pumps uere tried on the 4 K system uith the ultimate
selection of a Goulds MB 13600 stainless steel-ceramic, multi-stage,
centrifugal pump uhich had been especially designed for reverse
osmosis applications. For field applications, the 4 K unit uas
mounted in a portable 2.44 m x 6.10 m (8 ft x 20 ft) metal building.
In addition to tests at Norton, the studies at Morgantoun, Ebensburg,
and Mocanaqua uere made on this 4 K unit.
For the Mocanaqua study, a hollou-fiber unit manufactured by
DuPont uas tested alongside the spiral-uound 4 K system. Initial
11
-------
Mocanaqua hollnuj-fiber tests were made on a single permeator on loan
to Rex Chainbelt from the manufacturer. EPA later purchased tun
additional permeators to enlarge the hollou-fiber system to 6 K
(22,710 I/day or 6,ODD gpd of product flow) in order to more
effectively compare hollow-fiber results with spiral-wound.
Approximately 139.6 m2 (1,500 ft2) of B-9 modified nylon hollow-
fiber membrane were packed in each 15.2 cm x 1.22 m (6 in x U ft)
stainless steel pressure vessel as illustrated in Figure 3. The
three permeators were arranged in a 2-1 array. System arrangement
will be more thoroughly discussed in the section on the Mocanaqua
testing. All flows were measured volumetrically.
ANALYTICAL PROCEDURES
7
Acidity was determined according to the Salotto method of adding
hydrogen peroxide and titrating potentiometrically to a pH 7.3 end
point. Total iron, calcium, magnesium, manganese, and aluminum con-
o
centrations were determined by atomic absorption spectrophotometry.
n
EPA Methods was used for the sulfate determination. The pH was
measured potentiometrically. Ferrous iron was determined by color-
imetrically titrating potassium dichromate against a p-Diphenylamine
g
sulfonic acid sodium salt indicator- A YSI Model 51 dissolved
oxygen meter was used for dissolved oxygen measurements.
For field studies, acidity, ferrous iron, pH, conductivity, and
dissolved oxygen determinations were made on-site. Acidified samples
were returned to Norton for the remainder of analyses.
CALCULATIONS
Recovery was defined as the percentage of water resulting as
product as compared to the initial volume entering the unit. There-
fore, recovery equalled product flow 7 feed flow expressed as a per-
centage. Since accurate measurements could be made of the brine and
product flow rates, feed flow was calculated as the sum of the brine
and product flow.
12
-------
Snap Ring
Flow Screen
Open End
of Fibers
Ep ox y
Tube Sheet
Porous
Back-up Disc
Snap Ring
PERMEATE
End Plate
CONCENTRATE
Fib
'O' Ring
Shell Porous Feed Seal
Distributor Tube End Plate
e r
Figure 3
Hollow-fiber reverse osmosis module' '
-------
Feed = Product + Brine
RD Unit
Product
Brine
Since very little of the original pollutants appear in the prod-
uct stream, virtually all remain in the brine stream. As recovery
increases, progressively less water is available for dilution of
these pollutants. Figure k illustrates the recovery and concentra-
tion factor relationship. Above 75 percent recovery, small changes
in recovery level produce large changes in brine concentration.
The standard way of expressing membrane performance is flux rate;
i.e., gallons of product flow per square foot of membrane per day at
a specified temperature and net pressure. To obtain these values,
product flow was corrected from the observed temperature to 77° F
by a correction chart supplied by each manufacturer. The 77° F
temperature was used since it is the standard of the reverse osmosis
industry; however, a more realistic value for mine drainage studies
would be 50° F. In general, product flow increased roughly 1.5 - 2
percent per degree Fahrenheit. Factors for correcting flux rates
from English to metric units are supplied at footnotes for Figures
and Tables in this report.
Average pressure was determined by reading gage pressures in and
out of each vessel and averaging them. Osmotic pressure, which in
this study was mathematically correlated to conductivity, was sub-
tracted from the average applied pressure to give net driving pres-
sure.
Therefore, flux rates were determined according to the following
equation:
2
Flux (gal/ft /day @ 77° F and desired net pressure) = Prod-
uct flow (gpm) at observed temperature x correction factor
to 77° F '- (Average pressure - osmotic pressure) psi net ~
2
membrane area (ft ) x 1^0 minutes/day x desired net pressure.
-------
25r
-o
o
20
QC
O
15
<
ex.
V
Z
O
10
20 40 60
RECOVERY, percent
80
100
Figure 4. Effect of recovery on brine concentrations.
-------
Salt rejection measures the ability of a membrane to reject
specific ions; therefore, rejection is a measure of the efficiency
of a membrane in separating pollutants from the product water by
being relatively impermeable to the passage of salts. This measure-
ment is generally made in reference to ion concentrations in the
water entering the reverse osmosis unit as compared to the treated
water discharged. For these studies, percent salt rejection equalled
(concentration of influent - concentration of product) x 1DO/ concen-
tration of influent. In the case of the hollow-fiber unit at
Mocanaqua, the influent was raw acid mine drainage. For all spiral-
wound studies, blended feed was the influent to the unit.
Loss in flux has been observed on RD membranes which were treat-
ing high purity water. This flux loss phenomena in pure water sys-
tems is due to compaction of the membrane and/or its porous support
structure. ' In systems free of fouling, this flux loss follows
a linear log-log plot in respect to time. Under actual treatment
conditions, some fouling occurs in every application and the flux
trend may approximate a straight log-log line for varying lengths
of time until a significant buildup of fouling occurs. At that
time, an obvious deviation occurs as the flux line begins to curve
downward. Also, the slopes of flux loss trends are steeper than
for compaction losses alone.
The log-log slope is very important in analysis of reverse
osmosis operation as it is an indicator of the severity of fouling
and a rough predictor of membrane life. Figure 5 illustrates the
effect of various log-log slopes upon flux in respect to time.
Flux decline slopes in excess of -0.05 are felt to be intolerable.10
Therefore, membrane performance, deterioration, and fouling were
evaluated by platting observed flux values on log-log paper and cal-
culating log-log flux decline slopes from the line of best fit.
Taking values from the graph, the slope was calculated by the fol-
lowing equation:
Log Flux2 - Log
Slope =
Log Time? - Log Time,
16
-------
10%
200
8 1000
2 4
TIME, hours
81 10,000
1
. 4,
345
YEARS
Figure 5. Effect of log-log slopes on flux (all fluxes were 100% @ 1 hour).
-------
SECTION V
RESULTS
MAXIMUM RECOVERY STUDIES, REVERSE OSMOSIS AND NEUTROLOSIS
As previously mentioned, the main thrust of EPA's 1969 contract
with Gulf Environmental Systems was toward increasing recoveries in
order to produce the smallest possible volume of waste brine. These
studies were conducted at EPA's Norton Mine Drainage Field Site at
Norton, West Virginia.
The acid mater source for these studies uas Grassy Run, a sur-
face stream of which 90 percent of the floiu emanated from abandoned
coal mines. Due to surface flow and biological influence, the fer-
rous iron was rapidly oxidized to ferric. By the time the water
reached the Norton Facility, more than 90 percent of the iron was
in the ferric state. As a surface flow, wide variations in tempera-
ture were observed in relatively short periods of time. In addition
to the AMD, Grassy Run also contained sewage from approximately 30
residences upstream of the EPA Facility.
Typical water quality of Grassy Run during 1969 is presented in
Table 1.
The 10 H spiral-wound reverse osmosis unit was used for these
tests. As shown in Figure 2, sand filtration and 10 micron car-
tridge filtration served as pretreatment for the RO unit. Osmotic
2
pressure was assumed to equal 70 kN/m (10 psi) per 1000 micromhos
conductance.
Five tests were made during the high recovery stage of RQ
testing:
Test Number One: A 100 hour, 91 percent recovery reverse
osmosis run. This test was monitored 24 hours per day.
IB
-------
Table 1. TYPICAL GRASSY RUN WATER QUALITY - 1969
Parameter
pH
Specific
Conductance
Acidity as
CaC03
Calcium
Magnesium
Aluminum
Sulfate
Iron (Total)
Temperature
Flow
Units
pH
Mmhos/cm
mg/1
mg/1
mg/1
mg/1
mg/1
mg/1
°F
CFS
Mean
2.8(a)
1300
550
120
33
32
690
120
51
k.5
Max-
imum
3.1
1600
1700
170
52
50
1100
330
66
13.0
Min- Standard
imum Deviation
2.
-------
Test Number Two: A 96.5 hour reverse osmosis test in uhich
the recovery uas maintained at 91 percent during the day (7 hr/day)
and uas lowered to 85 percent during the night (17 hr/day). This
test uas monitored only during the normal 8 hr uork day.
Test Number Three: A 29.5 hour feasibility test of the
neutrolosis system. In neutrolosis, the brine uas lime neutralized,
the solids were removed, and the supernatant uater uas recycled into
the feed to the reverse osmosis unit. The results of this test were
previously reported11 and indicated that further tests of neutrolosis
should be made.
Test Number Four: A 99.6 hour neutrolosis test uith the
reverse osmosis unit operating at a recovery rate of 91 percent.
The recovery rate of the neutrolosis system was 98.8 percent. The
test uas monitored around the clock.
x
Test Number Five: A 130.6 hour neutrolosis test uith the
•
reverse osmosis recovery rate lowered to 80 percent. The neutrolosis
system recovery rate uas 98.7 percent and the test uas monitored con-
tinuously.
Test Number One (Reverse Osmosis)
This 100 hr reverse osmosis test uas made at 91 percent water
recovery. The operational and chemical data are given in Tables 2
and 3. Salt rejections uere uniformly above 99 percent on all multi-
valent ions. Figure 6 presents the history of membrane performance
during the test run. Both the unit flux and the tube 5 flux values
decreased uith time as the run progressed due to membrane fouling.
An explanation for the up and down flux variation is not available.
The tube 5 fouling rate uas somewhat more severe than that for the
total unit as tube 5 uas the last one and uas subject to the highest
ion concentrations. A greater risk of precipitation would be ex-
pected. Calcium sulfate precipitation appeared to be the predomi-
nant fouling mechanism as CaSO, crystals formed in brine samples
uithin one hour after sampling. The rate of fouling uas severe
20
-------
Table 2. OPERATING PARAMETERS FOR 10K NORTON RO STUDY AT 91 PERCENT
RECOVERY, TEST NUMBER ONE
Parameter Unit
Raw Water Feed Flow gpm
Product Water Flow "
Brine Water Discharged "
Brine Water Recycled "
Minimum Brine/Product Flow Ratio ratio/module
Maximum Brine/Product Flow Ratio "
Water Recovery percent
Recovery of Blended Feed "
Feed Pressure psi
Feed Temperature °F
Value
5.75
5.2k
0.51
3.69
9:1
17:1
91.2
55.5
600.6
kk
Unit Flux, gal/ft/day §
600 psi & 77° F
Tube Five Flux, gal/ft2/day §
600 psi & 77° F
20.50
19.55
Length of Run
Date of Run
hours 100.2
March 9-13, 1970
All l/alues are Means from 26 Data Seta
* Last tube in unit
Note: To convert flux (gal/ft2/day § 600 psi and 77° F) to liters/
2 2
m /day @ ^138 kN/m and 25° Cf multiply by **0.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.1^5.
21
-------
Table 3. CHEMISTRY ANALYSES FOR TEST NUMBER ONE
Rau Feed
Blended Feed
Brine
Product
(a)
Rejections
All units are mg/1
t-)
va'Bo ^ofUnn Fn.ialc
Cond-
uctance
1200
4200
9500
250
94.1%
pH
2.7
2.2
2.0
3.4
Acid-
ity
630
2600
5900
115
95.5%
Calc-
ium
110
480
1100
3.0
99.4%
Magne-
sium
33
180
410
0.7
99.6%
Alum- Total
inum Iron
35 110
170 530
400 1200
1.1 2.8
99.4% 99.5%
Sul-
fates
810
4000
9500
17
99.6%
except pH and conductance (Mmhos/cm)
Blended Feed
Concentration-Product
Concentration
X 100.
Blended Feed Concentration
-------
ro
ui
24.0
22.0
O
o
20.0
18.0
- 16.0
D
O)
X
= 14.0
LL.
C*
LU
t—
<
-I 1000
0
© UNIT PERFORMANCE
a TUBE 5 PERFORMANCE
900
CN
E
800 ro
-a
c
o
u
700
cs
o
T3
E
600 ^
- 500
20
40
120
140
60 80 100
HOURS OF OPERATION
Figure Number 6
Membrane performance during R.O. Test Number One at 91 percent recovery
-------
enough that once per week flushing would be required to maintain
membrane performance. Long-term operation of this system uith
the incorporation of flushes appeared possible.
Test Number Tuo (Reverse Osmosis)
During this 96.5 hr reverse osmosis test run, the unit operated
at 91 percent recovery during the day (7 hr/day) and was lowered
to 85 percent recovery at night (17 hr/day). Table *t presents a
comparison of typical operational data at the two recovery rates.
Flux values were higher at 85 percent than at 91 percent recov-
ery due to less fouling at the lower recovery.
In Table 5, the chemical analyses show that the salt rejections
were virtually identical at both recovery rates and were comparable
to the rejections of Test Number Dne. The concentration of pollut-
ants in the brine at 91 percent recovery was roughly 1.2 times the
respective concentrations at 85 percent recovery.
As lowering the recovery tended to reduce the fouling, a test
was made to determine if the fouling mechanisms could be detected
by chemical analyses and subsequent material balances. During the
run, samples were taken immediately before and 30 min after lowering
the recovery from 91 to 85 percent. At 91 percent recovery, the
quantity of all constituents entering the unit agreed within 10 per-
cent with the quantity leaving the unit. When the recovery was
lowered to 85 percent, the mass balance indicated approximately 30
percent more of each ion (calcium, sulfate, iron, aluminum, and
magnesium) was coming out of the unit than was going in. These
results clearly showed that the in-unit precipitate consisted of
more than calcium sulfate alone and that the precipitate could be
at least partially flushed from the unit by lowering recovery and
increasing brine flow.
A spectrochemical analysis of material scraped from the surface
of a membrane used in another test was reported by Sleigh to be
predominately chromium (from the high pressure pump), copper (from
-------
Table 4. OPERATING PARAMETERS FOR SPLIT RECOVERY REVERSE OSMOSIS
TEST (TEST NUMBER TWO)
Parameter
Hours per Day
Rau Feed Ulater Flow, gpm
Product Flou, gpm
Brine Flow Tube Five, gpm
Brine Flow Discharged, gpm
Brine Flou Recycled, gpm
Ulater Recovery Percent
Recovery of Blended Feed, Percent
Feed Pressure, psi
Feed Temperature, °F
Unit Flux, gal/ft2/day @ 600
psi Net & 77° F
Tube Five Flux, gal/ft2/day §
600 psi Net & 77° F
91 Percent
Recovery
7
5.1*2
4.94
4.09
0.48
3.61
91.1
54.7
600.0
45°
18.32
16.80
85 Percent
Recovery
17
5.63
4.94
4.11
0.69
3.42
84.4
54.6
602.5
44°
18.79
17.86
Length of Run, Hours
Date of Run
96.5
March 24-27, 1970
All Values are Means from Six Data Sets
Note: To convert flux (gal/ft2/day § 600 psi and 77° F) to liters/
m2/day @ 4138 kN/m and 25° C, multiply by 4QB67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
25
-------
Table 5. CHEMISTRY ANALYSES - SPLIT RECOVERY 10K REVERSE OSMOSIS
ro
en
Raw feed
Blended feed
Brine
Product
Rejections(a)
Rau feed
Blended feed
Brine
Product
Rejections(a)
All units are mg/1
Con-
ductance
1100
3600
7000
240
93.3%
1200
3000
6000
240
92.0%
pH
2.8
2.5
2.3
3.1
2.7
2.5
2.3
3.1
Acid-
itv
490
2000
4300
40
98.0%
540
1700
3900
37
98.0%
except conductance (Mmhos/cm)
. Blended Feed
Cal- Magne-
cium slum
90.7% Recovery
100 37
500 190
1050 440
4.0 1.0
99.2% 99.5%
84.8% Recovery
110 37
410 160
890 380
3.2 0.7
99.2% 99.5%
and pH.
Alum-
inum
31
160
360
1.0
99.4%
31
130
290
1.0
99.2%
Total
Iron
80
380
860
2.0
99.5%
80
320
680
2.0
99.4%
Sul-
fates
820
2600
8500
14
99.5%
850
1800
7100
12
99.3%
Concentration-Product Concentration y mn
Blended Feed Concentration
-------
bronze fittings), sodium (from WaCl test solution), silicon (from
pump packing), and iron. Iron uas the only element that could be
directly attributed to the acid mine water. This module had been
thoroughly acid washed before the analysis ujas made, and it must
be assumed that the absence of calcium and sulfur on the membrane
implies that CaSD, can be totally removed by prolonged flushing
whereas iron cannot. Possibly iron deposition may prove to be the
long-term threat to flux. However, calcium sulfate precipitation
is the immediate threat as massive CaSD, precipitation uas observed
in high recovery brine samples within one hour of sampling.
Test Number Three (Neutrolosis)
Brine disposal from a.n RD system is a severe problem. Even at
91 percent recovery, 9 percent of the water treated resulted as a
highly polluted liquid brine. Because of CaSO, fouling, 91 percent
uas near the maximum attainable recovery under the test conditions
at (Morton.
Previously, the Morton Mine Drainage Field Site had conducted
12
extensive research on neutralizing acid uater and RO brine.
Originally, the plan for controlling brine was to lime neutralize
the brine at 91 percent recovery and then blend the neutralized
supernatant with the product water from the reverse osmosis unit,
fully realizing that the neutralized supernatant would degrade the
high quality product water.
The idea of "neutrolosis" uas then conceived. Instead of blend-
ing the neutralized brine supernatant with the product, blend the
neutralized supernatant back into the feed to the RD unit. In this
manner, the only effluents from the process would be RO product of
a high quality and a neutralized sludge. The major concern was
that the higher pH of the neutralized supernatant would increase
the pH of the blended feed to a point where ferric iron would
hydrolyze and precipitate (around pH k).
27
-------
Test Number Three was an initial 3D hour feasibility test of the
neutrolosis concept. As the results of this test have been previously
reported,11 it uill suffice to mention that the reverse osmosis unit
recovery was 91 percent, the brine was neutralized, and the super-
natant did not raise the blended feed pH enough to cause iron precipi-
tation and the overall neutrolosis water recovery was 98.3 percent.
The apparent success of this feasibility test showed that the
neutrolosis concept had great potential and that further tests were
in order.
Test Number Four (Neutrolosisj
The neutrolosis concept was used for this test. A complete flow
diagram of neutrolosis (reverse osmosis-neutralization) is presented
in Figure 7. For the neutrolosis test, brine from the RD unit passed
directly into a 169 liter (50 gal) stainless steel reaction tank where
lime was added. The neutralized brine then was pumped from the reac-
tor to a 45^2 liter (12DD gal) upflow settling tank. In the settling
tank, the majority of the solids mas removed as sludge and the super-
natant was filtered and returned to the feed to the RD unit. Thus,
the only effluents from the process were (1) RD product water and
(2) neutralized brine sludge.
For this 99.6 hr test, the RD unit operated at 91 percent recov-
ery. Total recovery was 99.D percent. Table 6 presents the opera-
tional data for the test run and Table 7 presents the chemical
analyses.
As in Tests Dne and Two, the salt rejections in Test Number Four
were better than 99 percent on all multivalent ions. Lime neutrali-
zation, as seen in Table 7, removed virtually all the iron from the .
brine. However, pH k.l was not high enough to accomplish total
aluminum removal and 460 mg/1 of acidity remained in the supernatant.
Even though calcium (lime) was added, there was a net loss in calcium
and in sulfate indicating that a significant amount of calcium sulfate
was removed during the neutralization process.
28
-------
REVERSE OSMOSIS UNIT
PRODUCT
I
10 MICRON)
vCARTRIDGEl
HIGH
PRESSURE
MINE
WATER
CARTRIDGE
FILTER
SAND
FILTER
BRINE
RECYCLE BRINE
O!
UPFLOW
SETTLING
TANK
[1200 GAL)
'(4540
LITER)
LIME
FEEDER
pH METER
REACTION
TANK
(50 GAL) MIXER
(189 LITER)
SLUDGE'
Figure 7
Flow diagram for neutrolosis tests
PRODUCT
-------
Table 6. OPERATING PARAMETERS* FDR NEUTROLOSIS STUDY AT 91 PERCENT
REVERSE OSMOSIS RECOVERY (TEST NUMBER FOUR).
Raw water feed flow
Product water flow
Brine water discharged
(flow to neutralizer)
RT*T no i.ia+'OT» T*or»i/r?l ori
Unit
gpm
••
n
n
Value
5.02
4.97
0.48
3.49
ratio/module
gpm
Minimum brine/product
flow ratio
Maximum brine/product
flow ratio
Neutralized brine supernatant
recycled
Neutralized brine sludge
discharged
Reverse osmosis water
recovery
Recovery of blended feed
Total system recovery -
neutrolosis "
Percent sludge (by volume
of discharged brine) "
Feed pressure psig
Feed temperature °F
Unit flux, gal/ft2/day @ 600 psi net and 77°F
2/J m 600 psi net and 77°F
percent
n
Tube five flux, gal/ft /day
Length of run
Date of run
hours
6.3:1
24:1
0.43
0.05
91.2
55.6
99.0
10.4
602.2
58
15.43
9.83
99.6
May 11-15, 1970
*A11 values are means from 82 data sets.
,2
Note: To convert flux (gal/ft2/day @ 600 psi and 77°F) to liters/
2 2
m /day § 4138 kN/m and 25° C, multiply by 40.67; to convert gallons
per minute to liters/minute, multiply by 3.785; and to convert psi
to kN/m2, divide by 0.145.
3D
-------
Table 7. CHEMISTRY ANALYSES FOR NEUTROLOSIS TEST AT 91 PERCENT RO RECOVERY
(TEST NUMBER FOUR)
Rau Feed
Blended Feed
Brine
Neutralized Brine
Recycled
Product
Rejections
R. 0. Recovery at
Cond..
Mmhos/cm
1700
5200
10000
3700
340
93.5%
91 Percent
pH
2.7
2.2
2.0
4.7
3.4
Acid-
ity
660
2700
6000
460
130
95.196
Cal-
cium
100
470
1100
760
3.0
99.4%
Magne-
sium
38
270
550
310
1.5
99.4%
Alum-
inum
38
190
410
65
1.1
99.4%
Total
Iron
120
520
1200
1.5
1.6
99.7%
Sul- Alk. as
fates CaCO-j
980 0
4700 0
11000 0
2200 1.0
21 0
99.6%
All units are mg/1
Rejection equals
except conductance (Mmhos/cm) and
Blended feed
concentration-product
PH
concentration Y 1nn
blended feed concentration
-------
Comparing the product quality of this neutrolosis test with that
of Test Number One, (91 percent recovery RD test, Table 3) the prod-
uct qualities mere virtually identical. The only effects of recy-
cling the neutralized supernatant uere seen in the blended feed
concentrations. The neutrolosis blended feed had increased concen-
trations of magnesium and aluminum due to insufficient removal
12
during neutralization. Previous studies indicated both these con-
stituents could be removed around pH 7. If the magnesium and alum-
inum are not removed, a long-term buildup mould occur and neutrali-
zation should be taken close to pH 7 to alleviate this problem. Any
ions unich uiould not precipitate might be removed by occasionally
increasing pH and "blaming damn" the system.
Figure 8 illustrates membrane performance during the run. The
sharp decrease in tube 5 flux in respect to the unit flux indicated
severity of the fouling. Calcium sulfate precipitation was sub-
stantially morse in tube 5 than in the rest of the unit as tube 5
was subject to the heaviest concentration of pollutants.
Four'lorn-recovery (50 percent) flushes uere required during the
first 90 hr of the test to remove the severe fouling. Flush number
2
five uas a 3 hr flush at lou pressure (690 kIM/m or 100 psi) and lou
recovery (itO percent) using BIZ enzyme detergent. The BIZ flush uas
more successful in increasing flux than mere lou recovery acid flushes,
The severe fouling observed in Test Number Four mould prohibit
feasible operation under these conditions.
Test Number Five (Meutrolosis)
By operating the RO unit near 80 percent recovery and employing
the neutrolosis concept, it mas hoped that the severe fouling ob-
served in Test Number Four (neutrolosis test at 91 percent recovery)
could be reduced to tolerable levels. As louering the RO recovery
increased the volume of brine discharged, the flom of neutralized
supernatant also increased and again the possibility of the super-
natant increasing the blended feed pH to the point mhere ferric iron
mould precipitate mas feared.
32
-------
S 24.0
c
o
o
•o
-o20.0
c
o
16.0
CM
* 12.0
o
O)
£ 8.0
ot
4.0
0
Figure
es
It
CO
m
*
QUnit Performance
QTube 5 Performance
1000
800
,£
Z
CO
CO
•o
c
o
u
o
IO
X
o
•o
\
es
400 .5
600
200
20
100
120
40 60 80
HOURS OF OPERATION
8. Membrane performance during neutrolosis test (Number Four)
operating at 91 percent unit recovery and 98.8 percent system recovery*
-------
Table 8 presents operational data for the run and Table 9 pre-
sents chemical data. The recycled neutralized brine flow was
4.731 1/m (1.25 gpm) at 80 percent RD recovery as compared to
1.628 1/m (0.43 gpm) in Test Number Four (91 percent RO recovery).
However, the pH of the blended feed was still only pH 2.5, far from
the danger of ferric iron precipitation.
Total recovery during this 130.6 hr test uas 98.8 percent at 78
percent RO recovery as compared to 99.0 percent recovery (91 percent
RO recovery) in Test Number Four- Figure 9 shows the membrane per-
formance for the test. The rate of flux decline was significantly
less at 80 percent RO iiecovery (Test Five) as compared to Test Four
(91 percent RO recovery). Both the unit flux values and the tube 5
flux values declined at approximately equal rates during this 80 per-
cent test, indicating the fouling was uniformly distributed through-
out the unit. In the ejarlier tests, the tube 5 fouling rate was
noticeably more severe than the unit fouling rate.
Comparing this neutrolosis test at 80 percent RO recovery with
Test Number One (reverse osmosis test at 91 percent recovery -
Figure 5), the fouling rates were almost identical.
As in the previous tests, salt rejections were near 99 percent
on all multivalent ions.
A 2.5 hr low-recovery (50 percent) flush improved the flux rate
7 percent for the unit and 38 percent for tube 5, indicating that
the majority of earlier fouling had occured in tube 5.
Operation of this neutrolosis system at 80 percent RO recovery
and 98.8 percent total recovery appeared feasible for long-term use
if periodic flushes were incorporated to control fouling and sustain •
membrane performance.
Piscussion of Five Maximum Recovery Tests
The chemical analyses of the blended feed (water actually entering
RO unit) and brine were compared in an attempt to determine the cause
of Test Four's severe fouling as compared with Tests One and Five.
3k
-------
Table 8. OPERATING PARAMETERS* FOR NEUTROLOSIS STUDY @ 78 PERCENT
RO RECOVERY (TEST NUMBER FIVE)
Parameter
Raw Water Feed Flow
Product Water Flow
Brine Water Discharged (Flow
to Neutralize!-)
Brine Water Recycled
Minimum Brine/Product Flow Ratio
Maximum Brine/Product Flow Ratio
Neutralized Brine Supernatant Recycled
Neutralized Brine Sludge Discharged
Reverse Osmosis Water Recovery
Recovery of Blended Feed
Total System Recovery-Neutrolosis
Percent Sludge (by Volume of
Discharged Brine)
Feed Pressure
Feed Temperature
Unit Flux, gal/Ft2/day at 600 psi Net &
TubR 5 Flux " " "
Unit
gpm
it
n
ii
ratio/module
n
gpm
n
percent
n
n
n
psig
°F
77" F
Value
4.83
4.77
1.31
2.65
6:1
23:1
1.25
0.06
78.5
54.6
98.8
4.6
600.0
61.5
12.28
9.99
Length of Run hours 130.6
Date of Run June 3-8, 1970
*A11 values are means from 34 data sets.
Note: To convert flux (gal/ft2/day @ 600 psi and 77° F) to liters/
m2/day @ 4138 kN/m2 and 25° C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
35
-------
Table 9. CHEMISTRY ANALYSES FDR NEUTROLOSIS TEST AT 7896 RO RECOVERY (TEST NUMBER FIVE)
Raw feed
Blended feed
Brine
Neutralized
recycled
Product
fa
Rejections
pH Cond.
2.7 1660
2.5 3000
2.3 5600
brine
4.9 2500
3.7 210
} 92.9%
Acidity
590
1150
2400
180
36
96.9%
Calcium
130
460
940
850
2.1
99.5%
Magnesium
32
120
250
130
0.6
99.4%
Aluminum
43
91
190
30
1.0
98.9%
Total Alk. as
Iron Sulfate CaCO-j
100 1340 0
210 3000 0
460 6400 0
7.7 3500 7
1.2 18 0
99.4% 99.4%
All units are mg/1 except conductance
(a)
VH/Re.ior«f ion
, Blended
(Mmhos/cm)
and pH.
feed concentration-product concentration .. lnn
Blended feed concentration
-------
- 16.0
9
C
o
o
o 14.0
TJ
c
o
o
K
^ 12.0
X
O
•o
C1
10.0
o
O)
X~
3
i 8.0
oc
UJ
t—
<
0Unit Performance
QTube 5 Performance
600
E
"v.
Z
ao
CO
TJ
C
o
o
•o
CM
500
o
•o
^^
es
E
400 .^
ae
UJ
300 <
20
40 60 80
HOURS OF OPERATION
100
120
Figure 9. Membrane performance during neutrolosis test (Number
Five) operating at 78 percent unit recovery and 98.7
percent system recovery.
-------
These analyses, as shown in Table ID, indicated that the water
characteristics were identical between Tests One and Four, except
for magnesium and sulfate concentrations which were slightly higher
in Test Four, yet only Test Four fouled severely. All constituents
were much lower in concentration in Test Five than in either Test
Four or Test Dns as the reverse osmosis unit recovery was only 78
percent as compared to 91 percent on the other two tests. Still,
Test Five fouled at the same rate as Test Dne.
The conclusions drawn from these data were that the fouling in
all tests was due to precipitation, but this precipitation apparently
was not directly related to chemical concentration (as the concen-
trations in Test Five were significantly less than Test Dne, yet the
fouling rates were similar; and the concentrations were approximately
equal in Tests Four and One, yet Test Four fouled severely). Instead,
the physical properties of the recycled neutralized brine supernatant
appeared to promote precipitation. Since the neutralized recycle
was already above saturation for calcium sulfate, it was assumed that
minute CaSO, seed crystals existed in the recycled supernatant and
these promoted precipitation in the unit.
Product water quality, as shown in Table ID, was much superior in
Test Five as compared to Tests One or Four as the lower RD unit re-
covery (78 percent) did not concentrate the pollutants as much as
the higher recovery (91 percent) runs. Since salt rejections were
approximately the same in all these tests, lowering the brine con-
centrations improved product quality. f\)one of the product waters
were of potable quality. Further treatment with lime to increase
the pH and remove the iron would be required to increase the quality
to potable standards.
In conclusion, the feasibility of 91 percent recovery reverse
osmosis operation under the conditions at Norton was borderline. If
periodic flushing would continue to control fouling, operation would
be feasible. Disposal of the 9 percent highly polluted liquid brine
would still be a major problem.
38
-------
Table 10. BLENDED FEED, BRINE AND PRODUCT CHEMISTRY ANALYSES FOR 10K STUDIES
vfl
Test
HI
Reverse osmosis
#4
Neutrolosis
#5
Neutrolosis
Reverse osmosis
#4
Neutrolosis
#5
Neutroloais
Reverse osmosis
Neutrolosis
#5
Neutrolosis
RO
recovery
91.2
91.2
78.5
91.2
91.2
78.5
91.2
91.2
78.5
Total
recovery
91.2
99.0
98.8
91.2
99.0
98.8
91.2
99.0
98.8
Cond. pH
BLENDED FEED
4200 2.2
5200 2.2
3000 2.5
BRINE
9540 2.0
10,000 2.0
5600 2.3
PRODUCT
250 3.4
340 3.4
210 3.7
Acidity
2580
2700
1150
5900
6000
2400
115
130
36
Ca
480
470
460
1100
1100
940
3.0
3.0
2.1
Mq
180
270
120
410
550
250
0.7
1.5
0.6
Al
170
190
91
400
410
190
1.1
1.1
1.0
Total
iron
530
520
210
1200
1200
460
2.8
1.6
1.2
Sulfate
4000
4700
3000
9500
11,000
6400
17
21
18
All units are mg/1 except for recovery (percent), conductance (Mmhoa/cm), and pH.
-------
The use of neutrolosis at 91 percent RD unit recovery produced
a total recovery around 99 percent and one percent of the initial
volume resulted as a dense sludge which could be landfilled. Unfor-
tunately, the fouling rate of the RO membranes under these conditions
was too severe for feasible long-term operation.
By employing the neutrolosis concept at 80 percent RO unit recov-
ery, fouling was minimized to levels where continuous long-term oper-
ation appeared feasible. Periodic flushes should be able to control
membrane fouling. Total recovery by this system was 98.7 percent.
The remaining 1.3 percent of the original water resulted as sludge.
The neutrolosis concept is a significant breakthrough in RO
technology. Converting 99 percent of the acid water treated into a
high guality product water and the remaining one percent into a dense
sludge is a dramatic result. The reverse osmosis problem of brine
disposal was eliminated in the Norton application by this neutrolosis
process. Ue see no reason why this process or modifications of it
could not be used to treat most types of acid mine drainage. Ferrous
iron sites may have to include aeration and higher pH neutralization
to remove the iron prior to recycling. At such sites, it may be
possible to inject acid to lower the pH of the neutralized brine
supernatant before recycling it into the feed. Sludge dewatering
devices would increase the recovery even further. Development of
calcium sulfate precipitation inhibitors would reduce membrane foul-
ing problems. In brief, neutrolosis is a very promising method for
the treatment of acid mine drainage.
TWO-STAGE REVERSE OSMOSIS SYSTEM
Gulf Environmental Systems had investigated several calcium sul-
fate precipitation inhibitors (chelating agents) which had been suc-
cessfully applied to sea water desalination3. If the precipitation
of CaSO^ could be delayed, then maximum recovery could be signifi-
cantly increased. Unfortunately, the chelating agents tested were
ineffective due to either acidic pH or high iron concentrations, or
both.
-------
In order to utilize sodium hexametaphosphate as a CaSO, precipi-
tation inhibitor, the water must first be neutralized. Neutraliza-
tion of the entire volume of uater to be treated by the reverse
osmosis unit was impractical. A more feasible scheme was to operate
the ID K RD unit on rau AMD at 91 percent recovery and to neutralize
the brine (9 percent of the original volume - therefore neutralization
costs would be less). After settling the solids, the neutralized
supernatant uater would be fed into a smaller, second, RO unit in
which sodium hexametaphosphate could be used as a chelating agent
since the pH would be close to neutral.
To test this arrangement, a k K unit, which was supplied under
the terms of a Gulf Environmental Systems Contract , was used as
the second-stage RO unit. The k H unit consisted of three 1.58 cm
x 3.05 m (*4 in x 10 ft) pressure vessels arranged in a 2-1 array
where the brines from tubes one and two combined to serve as the
feed to tube three.
Two tests were made of the proposed process. Brine from the 10 K
unit operating at 91 percent RO recovery was collected, neutralized,
and stored. From the storage tank, the neutralized water was pumped
into the k Y\ RO unit which operated at 50 percent recovery. A
schematic drawing of the process is presented in Figure 10; in
Table 11 are the data for the second test of the process. Overall
system recovery was 9^.75 percent. As seen in Table 11, a continual
decrease in product flow was observed during the 2.6 hr 4K run an
brine supernatant water which had been neutralized to pH k.Q. A
rapid increase in pressure loss across the unit occurred indicating
immediate and severe fouling and terminating the test. Sodium hex-
ametaphosphate was ineffective at pH it.8 in controlling CaSO pre-
cipitation. As very little iron was present in the neutralized feed
(Table 12), the inability of the chelating agent to perform success-
fully was probably the result of the acidic pH (i+.S). Mo further
tests were made at higher pH's, however-
Salt rejections during this test were unexplainably low (Table 12),
kl
-------
91% RECOVERY
AMD FEED
100% FLOW
10K R.O. UNIT
PRODUCT
BRINE
9% FLOW
I
NEUTRALIZATION
| SYSTEM |
(17% SLUDGE)
SLUDGE (@ 20%)
17% x 9% = 1.5 % FLOW
91% FLOW
83% x 9%= 7.5% FLOW
7.5% 50% RECOVERY
PRODUCT
4K R.O.
Unit
BRINE
1
SODIUM HEXAMETAPHOSPHATE
@ 10 ppm
Total Feed = 100%
Product Recovered
10K R.O. = 91% of 100%
4K R.O. = 3.75% of 100%
Total System Recovery = 94.75%
Waste
Sludge = 1.5%
4K Brine = 3.75%
f
3.75%
FLOW
3.75% FLOW
Figure 10. Two stage RO treatment with intermediate
brine neutralization.
-------
Table 11. OPERATING PARAMETERS FOR 50 PERCENT RECOVERY 4K RO STUDY ON NEUTRALIZED BRINE
Elapsed Product, Brine,
Time . hr qpm qpm
0
0,
0,
0,
0,
1,
1.
1,
1,
1,
2.
2.
2,
2.
.1
.1
.2
,2
.0
.1
.5
.6
.8
,1
.3
.5
.6
2.19
2.19
2.19
2.17
Samples taken
2.15
Samples taken
2.00
2.04
2.00
2.00
1.96
1.96
2
2
2
2
2
2
2
1
1
1
1
Unit shut down
.90
.63
.19
.17
.17
then
.11
.00
.94
.95
.92
.92
and
Recycle, Pressure,
qpm psi
2.50
2.50
3.00
3.00
3.25
pressure lowered
3.00
3.25
3.25
3.25
3.25
3.25
flushed with HP
410
410
410
415
420
400
400
400
400
400
400
1
AP^a; AP AP
Recovery System, Tubes 1 & 2, Tube 3,
psi psi psi
37.
45.
50.
50.
50.
51.
51.
48.
50.
50.
50.
50.
50.
1%
4%
0%
0%
0% 17.5 6.5
3%
3% 19.0 7.8
6%
4%
7% 20.3 8.5
6%
5% 22.1 9.4
5%
11
11
12
13
.3
.7
.2
.2
pH of influent = 4.8.
Temperature of water = 52° F (11° C).
Total operating time = 2 hr, 35 min.
Neutralized brine sludge volume = 17% (after 24 hours).
Total recovery (including 91% 10K recovery, 17% neutralized brine sludge, and 50% 4K recovery)
94.75% of original feed.
Sodium hexametaphosphate injected at 10 ppm rate.
Date of run - March 20, 1970.
(a)
AP = Pressure drop. „
To convert gpm to 1/s, multiply by 0.065, and to convert from psi to kl\l/m , multiply by 6.90.
-------
Table 12. CHEMISTRY ANALYSES FOR 50% RECOVERY 4K RO STUDY ON NEUTRALIZED 10K BRINE
Neutralized brine
Parameter (Feed to 4K unit)
pH
Conductivity
Acidity
Calcium
Magnesium
Iron
Aluminum
Sulfate
4.8
3200
66
400
170
1.3
4.5
3100
Blended feed
4.9
4000
73
540
210
1.6
4.8
4800
Brine
4.9
5000
110
800
300
2.0
8.0
6000
Product
4.5
200
3.9
10
4.0
0.1
0.3
60
(a;
Rejections
—
93.8%
94.1%
97.5%
97.7%
92.3%
93.3%
98.1%
All units are mg/1 except for pH and conductivity (micromhos/cm).
(a)
Rejection equals raw feed concentration-product concentration
ram feed concentration
X 100.
-------
MORGAMTOUM k H FERROUS IRDIM SPIRAL-LJOUND STUDY
Up to this time, the RD studies had been conducted an ferric
iron mine drainage. The next series of studies was conducted on
ferrous iron. The first ferrous site was the acid mine discharge
into Indian Creek located ID miles south of Morgantoun, West Virginia,
at the Arkwright Mine of Christopher Coal Company.
The k K unit and support wiring uere housed in a 2.k m x 6.1 m
(8 ft x 20 ft) metal building so that the entire system was portable.
Dn-site chemical analyses (acidity, ferrous iron, pH, and conductiv-
ity) were performed in this building and the samples then acidified
and returned to the Norton Laboratory for the remainder of the anal-
yses. Dissolved oxygen (D.D.) determinations, which were made at
the site, were unstable but in the ^.5 ppm range. Attempts at using
a modified Ldinkler Method on the brine were unsuccessful because of
the high ferrous iron concentration.
The flow diagram for the 4 H unit and support facilities is
shown in Figure 11. Neither brine recycle or pH control was used
during this study.
The 189D liter (500 gal) holding tank was necessary as the mine
water was pumped only intermittently from the borehole (on 8 minutes,
off 5 minutes). Pressured sand filters and 10 micron cartridge fil-
ters were used to remove any suspended solids from the feed water
before the water entered the RO unit.
Initially, tubes 1 and 2 contained standard-flux, high-selec-
tivity modules and tube 3 contained high-flux, intermediate-selec-
tivity modules. These modules had been in operation approximately
180 hr prior to this study. After the first test run, the modules
were badly fouled and were replaced. The loading for the second
test was: tube 1, new high-flux modules; tube 2, new standard-flux
modules; and tube 3, used high-flux modules.
The histories of the two Morgantown tests are presented in
Table 13 and the chemical analyses are reported in Table 1*4.
US
-------
138 kN/m2
PH SENSOR AND
CONTROLLER
(OPTIONAL)
en
172kN/m2/ SAND
10 MICRON
CARTRIDGE FILTERS
RECYCLE BRINE
4,38
(600 psi) max. RESERVE OSMOSIS UNIT
124kN/m2
(18 psi)
TUBE 1
MINE DISCHARGE
H TUBE 2
TUBE 3
BACK PRESSURE
REGULATOR
RESTRICTING VALVE I
CHEMICAL INJECTION
PUMP AND RESERVOIR
(OPTIONAL)
BRINE
276kN/mJ
(40 psi)
(500 gal)
Figure 11. Flow diagram for 4K spiral-wound reverse osmosis unit.
-------
Table 13. 4K REVERSE OSMOSIS OPERATIONAL HISTORY FOR MORGANTOUN FERROUS IRON STUDY AT THE
ARKUJRIGHT MINE
Date
Run #1
5/13/70
5/13/70
5/13/70
5/14/70
5/14/70
5/14/70
5/15/70
5/21/70
5/21/70
5/21/70
5/21/70
Run #2
5/22/70
5/22/70
5/22/70
5/23/70
5/24/70
5/24/70
5/25/70
5/25/70
5/26/70
Time,
hours
0
3.6
4.1
22.0
27.8
28.8
45.7
0
1.7
9.8
21.7
22.7
23.8
25.0
40.5
65.2
69.1
88.8
89.3
111.6
111.7
Elapsed Pres- Re- Product
time, sure, covery, flow,
hours osi percent opm
0 400 Unit started
3.6 400 39.5 2.30
4.1 500 47.0 3.33
22.0 500 43.5 2.86
27.8 600 49.1 3.57
Poiuer failure
Unit shut down - no feed pressure - pump
Modules badly fouled
Replacement modules installed (some used)
0 400 2.50
Unit shut down for inspection of tubes 1
1.7 Probed tubes for leaky modules
9.8 Shut down again for inspection
Replaced modules and 0-Rings
Brine
flow,
qpm
3.52
3.75
3.70
3.70
air lock
4.00
& 3 (high
AR(aJ Temp. Unit(b)
osi °F flux
9.9
11.0
11.0
11.1
cond. )
21.7 Feed pump leaking, unit flushed with acidified product
Replaced pump and 3 more modules
Restarted unit
0 400 41.6 2.64
1.1 600 52.5 3.87
2.3 600 51.4 3.84
17.8 600 50.0 3.45
42.5 600 47.5 3.23
46.4 600 46.7 3.13
66.1 600
Begin low-pressure flush
88.9 240 20.7 1.05
Terminate testinq
3.70
3.48
3.64
3.46
3.57
3.57
4.00 No
10.5
-
11.1
10.3
11.3
11.2
12.0
change
68
68
64
67
water
68.5
68
68
62
64
70
63
16.38
18.43
16.99
16.09
19.23
17.16
17.05
17.01
15.36
13.48
14.40
(a)
, .Pressure drop^across unit = pressure in - pressure out
*• Flux = gal/ft/day @ 77° F .and 600 psi net _ _
Note: To convert flux (gal/ft/day @ 600 psi and 77° F) to litera/ni /day @ 4138 kf\l/m and 25° C,
multiply by 40.67; to convert gallons per minute to liters/second, multiply by 0.063; and to con-
convert psi to kN/m2, divide by 0.145.
-------
Table 14. CHEMICAL ANALYSES FOR 4K REVERSE OSMOSIS STUDY AT ARK-
UIRIGHT FERROUS IRON SITE (UNIT RECOVERY 50 PERCENT)
Parameter
pH
Dissolved oxygen
Conductance
Acidity as CaCO,
Calcium
Magnesium
Aluminum
Sulfate
Iron (Ferrous)
Iron (Total)
% Ferrous
*Rejection = Feed
Feed
2.24
4.50
7000
5200
530
420
320
10,900
1300
2300
56.5%
Brine
2.00
4.50
12,000
10,000
930
810
600
20,500
2450
4460
54.9%
concentration-product
Product
3.14
-
420
150
9.6
7.6
5.0
190
29
39
74.4%
concentration
Rejections*
94.0%
97.1%
98.2%
98.2%
98.4%
98.3%
97.8%
98.3%
Y i nn
Feed concentration
All units are mg/1 except for conductance (micromhos) and pH.
-------
The first run, lasting U5.1 hr, was begun at low pressure and
L*Q percent water recovery. The recovery was gradually increased
by raining the pressure until 50 percent recovery mas established.
Unfortunately, data were available for only the first 27.8 hr of
operation. The differential pressure, which indicates rapid fouling,
increased initially but levelled off after four hours of operation.
However, performance of the membrane, as illustrated by the water
flux rate, showed a definite decreasing trend as shown in Table 15.
The second test was begun at 50 percent water recovery and lasted
66.1 hours, but data were available for only the first ^6.^4 hr. At
the end of 66.1 hr, the unit was badly fouled by precipitation which
occurred when a brass valve corroded and diminished the water supply
to the unit, thus the unit was receiving only recirculated brine.
As shown in Table 15, the water flux decreased 21 percent in 46.^ hr
of operation. The flux value for tube 3 (where the precipitation
problems are most acute due to the higher concentration of ions)
decreased only 19 percent. A 35 percent, flux loss occurred in tube
1, 5 percent in tube 2, and 19 percent in tube 3. The high flux
decrease in tube 1 (loaded with new high-flux modules) was difficult
to explain, especially since tube 2 was in parallel to tube 1 and
was subject to the same conditions. After termination of th° test
run, an explanation for the tube 1 flux loss was found. The brine
seal on the last module of the tube had failed and allowed water
to bypass the module. The module oroduceri progressively less water
until it became severely fouled by precipitates.
If results from tube 1 are ignored and flux declines between
tubes 2 and 3 are compared, the decrease appears to be a concentra-
tion related phenomena, as a much greater decrease was observed in
tube 3 where the concentration was the greatest.
There was no significant increase in the differential pressure
for the unit or for the individual tubes. This indicated that the
rate of fouling uas not too severe to prohibit short-tnrm operation
-------
Table 15. FLUX* AND AP VALUES FOR ARKWRIGHT FERROUS IRON STUDY
ui
a
Run
No.
1
2
1
2
Elapsed
Time
3.6
22.0
27.8
DIFFERENT
2.3
17.8
42.5
46.4
A p
3.6
22.0
27.8
2.3
17.8
42.5
46.4
Tube 1
Flux
14.58
14.75
13.55
MODULES
19.20
16.04
14.00
12.53
Tubes
1 & 2
(psi) &
3.4
4.0
4.3
4.0
3.5
3.5
3.5
Tube 2
Flux
14.58
15.31
14.23
14.00
15.24
14.57
13.32
Tube
3
. P (psi)
6.7
7.2
7.1
7.5
7.0
8.0
8.0
Tube 3
Flux
20.29
24.94
19.52
19.03
20.48
16.27
15.33
Unit
A P (psi)
9.9
11.0
11.1
11.1
10.3
11.3
11.2
Unit
Flux
16.38
16.99
16.09
17.05
17.01
15.36
13.48
*Flux = gal/ft /day at 77°F and 600 psi. Since the osmotic pressure of this water was unknown
and recovery was relatively constant, osmotic pressure was ignored for these calculations.
Note: To convert flux (gal/ft2/day @ 600 psi and 77° F) to Iiters/m2/day @ 4138 kN/m2 and
25° C, multiply by 40.67; and to convert psi to kflJ/m2, divide by 0.145.
-------
on this water at the 50 percent recovery level. Rejections of the
various ions by the membranes mere relatively high (98 percent on
all multivalent ions). Rejection of ferrous iron was somewhat lower
than the rejection of ferric iron as trivalent ions are better re-
jected than are divalent ions. Maximum brine/product (b/p) flow
ratios during the study were near 9.^:1. Minimum b/p ratios were
5.1:1.
Discussion of Morqantown Study
Chemical composition of the feed water at the Arkwright Site
was similar to that of the 91 percent recovery brine from the RD
unit at the (Morton Site (Table 3). Attempts to further treat the
brine from the Norton test by RD had revealed that calcium sulfate
fouling occurred at 50 percent recovery. Therefore, it was not
surprising that 50 percent recovery was the maximum attainable at
Arkwright.
The two test runs, though short-lived, were terminated by equip-
ment and facility problems, rather than module or membrane failure.
However, membrane fouling was evident. Fifty percent recovery was
about the maximum recovery possible before CaSO, deposition began
to rapidly foul the membranes.
Effects of fouling were easily seen during the study by the de-
crease in flux values for the membranes. Unfortunately, the exact
cause of fouling was difficult to determine. All evidence pointed
to calcium sulfate deposition, but the role of ferrous iron could
not be determined.
During the study, a sample of brine at 50 percent recovery was
collected and a precipitate formed within an hour. Analysis of the
precipitate (Table 16) indicated the composition to be virtually
CaSO,. Trace amounts of iron were present.
Lf
A sample of the precipitate that lined the pressure vessel was
analyzed (Table 17). The composition of the in-unit precipitate
was similar to the external brine precipitate (Table 16).
51
-------
Table 16. CHEMICAL ANALYSIS OF ARKWRIGHT STUDY BRINE PRECIPITATE
CD3
Fe
Mg
Sn,
-------
The majority of the precipitation in the unit occurred during
the last feu hours before the runs were terminated. In both test
runs, mechanical problems caused the recovery to soar to undetermined
levels (probably greater than 90 percent) before the unit shut down
and thus the brine was very concentrated and precipitation occurred.
As calcium sulfate is the first salt to exceed solubility, it was
expected that the precipitate would be largely calcium sulfate. The
results in Table 17 confirm this conclusion.
However, this predominance of CaSO deposition may not be repre-
sentative of what occurred during successful operation. The possi-
bility of ferrous iron fouling cannot be ruled out nor can it be
substantiated by these results.
Contrasting the salt rejections at Arkwright with those at Morton,
the rejections at Norton were slightly higher in most cases. This
difference in rejection rates was due to the high ionic concentration
at Arkwright whose higher osmotic pressure resulted in less product
flow while the salt flow remained essentially the same. This con-
clusion was valid since osmotic pressure is directly related to
ionic concentrations and increases in ionic concentrations elevate
osmotic pressure.
In conclusion, it does not appear that RO would be a feasible
method for treatment of water of this quality based upon the problem
of membrane fouling which severely limited recovery (50 percent)
resulting in disposal of an equal volume of brine for each gallon
of water reclaimed.
EBENSBURG U K FERROUS IRON SPIRAL-WOUND STUDY
A ferrous iron test site with concentrations more typical of AMD
commonly encountered was located near Ebensburg, Pennsylvania, at
the 32-33 discharge of Bethlehem Mines Corporation Cambria Division.
The k K spiral-wound test system was installed there in July 1970.
The study included tests to determine if iron fouling could be con-
trolled by pH adjustment, i.e., the lower the pH, the less liklihood
of iron precipitation.
53
-------
A new, improved version of Gulf Environmental System's standard
flux, high-selectivity module uas used in this Ebensburg study. In
these modules, the membrane uas cast directly upon the backing mate-
rial and the old glue joints uere eliminated. The glue joints may
have contributed to small "leakage" problems in the past and the
manufacturer felt that their elimination should insure high rejec-
tion rates.
The system flouj arrangement uas the same as that at Arkuright
(Figure 11) except for acid injection which uas used during this
study. Osmotic pressure uas assumed to equal ID psi per 1,000
micromhos of conductance, tilater quality of the Ebensburg site uas
similar to that at Morton (except for ferrous iron).
Four tests uere made during this study. These were:
Test Number One: A 191 hour, 80-85 percent recovery, recycle
brine run utilizing injection of sulfuric acid into the feed water to
louer the pH from pH 3.6 to pH 2.5. Acid injection uas used since it
uas feared that ferric iron might precipitate at pH 3.6. The amount
of acid injected uas gradually diminished touard the end of the test.
The last module in tube 3 uas removed at the end of the test and re-
turned to Gulf Environmental Systems for post mortem analysis.
Test Number Tuo: An 86 hr, 80-85 percent recovery, recycle
brine run with no pH control of the feed (i.e., no acid injection).
Test Number Three: An 18 hr, 80-85 percent recovery, recycle
brine run utilizing injection of sodium hydroxide to increase the pH
of the blended feed (blended feed = raw feed plus recycled brine;
blended feed is the uater actually entering the RO unit) to pH 3.6.
The purpose of this test uas to simulate the rau uater pH to deter-
mine if ferric iron precipitation uould occur in a unit uith no
brine recycle.
Test Number Four: A 3k hr, 50-55 percent recovery run uithout
brine recycle and no pH control. A neu module had been installed in
tube 3 at the start of this run. After the test, the module uas re-
-------
moved and returned to Gulf Environmental System for post mortem
analysis. This test investigated the fouling rate at louer re-
coveries.
Test Number One
As the pH of the AMD at Ebensburg was 3.6, the possibility of
ferric iron precipitation uas feared. Sulfuric acid uas, therefore,
injected automatically into the uater before it entered the RQ unit
in order to louer the pH to 2.5. The test run lasted 191 hr and the
amount of acid injected uas gradually diminished toward the end of
the run. The recovery during the test uas maintained near 8*4 per-
cent by recycling a portion of the brine into the feed to the unit.
As tube 3 ur-s subject to the most polluted water, it uas more
prone to fouling than tubes 1 and 2. As shoiun in Figure 12, the
pressure drop (AP) across tube 3 after 100 hr of operation had
increased significantly, thus indicating membrane fouling. The
recovery uas then reduced for three hours in an effort to remove
the fouling. The flush uas successful in restoring the AP across
tube 3 to normal values. Significantly, theAP across tubes 1 and
2 decreased during the run uhile tube 3 increased, indicating that
the-fouling uas a concentration-related phenomenon.
The flux history of the run is shoun on a linear basis in Fig-
ure 13. Very little fouling uas observed in tubes 1 or 2 as their
log-log slopes uere -0.0^3. An excessive flux decline uas observed
in tube 3. A rapid change in log-log slope for tube 3 occurred
after approximately ^0 hr of operation on the Ebensburg uater.
It is felt that the tube 3 flux loss mas due to calcium sulfate
fouling. Although acid injection uas diminished as the run pro-
gressed, very little change occurred in the tube 3 pH as shown in
Table 18. Thus, the decrease in acid injection probably had no
significant effect on tube 3 behavior.
55
-------
U1
en
75
O Tubes 1 & 2
Tube 3
50
25
50
150
100
ELAPSED TIME, hours
Figure 12. (AP) Pressure drop across tubes during Test Number One.
200
-------
0)
C
o
o
o
TJ 17.5
C
o
@;15.5
x
o
TJ
~ 13.5
o
u>
X
^ 11.5
9.5
o Tube 1
A Tube 2
D Tube 3
700
cs
E
00
to
TJ
C
O
u
o
CM
600
X
0)
500^
tt
LU
t—
<
- 400
50
100
ELAPSED TIME, hours
150
200
Figure 13. Membrane performance during Ebensburg test at 84 percent recovery.
-------
Table 18. BLENDED FEED AND BRINE pH'S AT EBENSBURG
Elapsed
time, hr
0-28
28-57
57-96
96-100
100-191
Blended
feed pH
2.7
2.75
2.8
3.2
3.1
Elapsed
time, hr
0-100
100-191
Brine
(tube
2.3
2.6
pH
3)
On the log-log graph, there appeared to be a slight tendency
for the tube 1 flux values to tail off from the linear log-log
slope of -0.0^3 between hours 100 and 191. Unfortunately, the run
did not continue long enough to substantiate this suspicion. If
some fouling uere occurring in tubes 1 and 2 during that time period,
it would likely have been due to iron precipitation as the pH (Table
18) had increased to 3.1 during that portion of the test.
A module mas removed from the downstream end of tube 3 at the
end of the test and returned to Gulf Environmental Systems for post
mortem analysis. Sleigh reported the presence of a white flaky
precipitate in the module:
"Analyses of this material showed it to be 23.7
percent calcium, 55.1 percent sulfate, and 22.6
percent loss of weight at 600° C. The theoret-
ical amount of CaSO^ in gypsum (CaSO,.2H 0) is
79.1 percent; therefore, this material was almost
pure gypsum. The precipitate could easily be
washed from the membrane which is what takes
place in the unit when the recovery level is
lowered."
No tests were made to determine the presence of iron.
Table 19 presents the physical data for the test run and Table
20 shows the chemical analyses. Salt rejections were uniformly
greater than 99 percent on all multivalent ions.
58
-------
Table 19. OPERATING PARAMETERS FOR TEST NUMBER ONE AT EBENSBURG, PENNSYLVANIA
m
Average recovery, percent
Average feed rate, gpm
Average brine flout rate, gpm
Average recycle brine flou rate, gpm
Average temperature, °F
Average feed pressure, psi
Average product flou rate, gpm
Normalized to 50°F, gpm
Normalized to 77 °F, gpm
83.6 Avg. 85.7 Max. 81.2 Min.
3.57
0.59
3.79
63.6 Avg. 67.5 Max. 59.0 Min.
399.2
2.98
2.32
3.72
2
Average flux rate, gal/ft/day @ 50°F and 400 psi net 7.44
Average flux rate, gal/ft2/day @ 77°F and 400 psi net 11.90
Maximum brine/product flou ratio 19.6:1
Minimum brine/product flou ratio 7.5:1
Dates of test : July 30 - August 7, 1970
Length of run : 191 hours
Number of shutdowns: 0
Note: To convert flux (gal/ft2/day @ 400 psi and 77°F) to Iiters/m2/day @ 2758 kN/m2 and 25° C,
multiply by 40.67; to convert gallons per minute to liters/second, multiply by 0.063; and to
2
convert psi to kN/m , divide by 0.145.
-------
Table 20. CHEMISTRY ANALYSES FOR TEST NUMBER ONE WHERE AVERAGE RECOVERY = 83.6%
en
o
pH
Conductivity
Acidity
Calcium as Ca
Magnesium as Mg
Aluminum
Total iron
Ferrous iron
Sul fates
Dissolved oxygen
Raw
feed
3.6
1500
380
190
5k
32
135
100
1640
2.0
Acidified
feed
3.0
1825
450
190
54
32
135
100
1620
—
Blended
feed
2.7
5100
1270
600
190
100
430
310
5100
•~
Brine
2.5
6900
2360
920
340
180
730
540
9300
2.0
Product
4.1
92
56
1.2
0.4
0.9
1.7
<2
14
2.0
Rejections*
98.2%
95.6%
99.8%
99.8%
99.1%
99.6%
99.4%
99.7%
~
All units are mg/1 except for conductivity (Mmhos/cm) and pH.
•Rejection equals
Blended feed concentration-product concentration
Blended feed concentration
X 100,
-------
Test Number Two
The injection of sulfuric acid to control the influent pH was
not used in this test as it was observed that, at 8Q-S5 percent
recovery, the recycled brine lowered the pH of the resulting "blended
feed" from pH 3.6 to pH 3.1.
This test was plagued with shutdowns. In a total run time of
86.3 hr, the longest continuous run was 22.5 hr and the longest
continuous period of data available was for only 17.3 hr. The shut-
downs were all caused by pump problems.
Table 21 shows the flux andAP value for Test Number Two.
Each accidental shutdown acted as a flux rejuvenator. Planned
full pressure shutdowns have long been used at Morton to dislodge
precipitates and restore membrane performance. In essence, each of
these accidental shutdowns served to clean and relax the membrane.
Considering the data between shutdowns as individual runs, the
flux values in most cases decreased as the run progressed. As in
Test Number One, the flux decline in tube 3 was significantly
greater than the decline in tubes 1 or 2.
The AP value over the entire test run did not change appreciably
for tubes 1 and 2 but did show a continuous increase across tube 3.
This was analogous to the results of Test Number One.
Flux rates for tubes 1 and 2 were greater at the end of the test
than at the beginning. The tube 3 flux decreased 9.2 percent over
the entire run.
Overall, there was significantly less fouling in Test Number
Two than in Test Number One, due to the shutdowns which acted as
precipitation removers.
The physical data for Test Number Two are shown in Table 22 and
the chemical analyses are given in Table 23. As in Test Number One,
the salt rejections of all multivalent ions were greater than 99
percent.
61
-------
Table 21. FLUX AND AP VALUES FOR EBENSBURG TEST NUMBER TWO
Elapsed
Time
(Hours)
Flux
Tube
1
Flux
Tube
2
Flux
Tube
3
AP
Tubes
1&2
AP
Tube
3
(psi)
(psi)
0.7
2.8
5.9
22.5
22.9
24.7
27.8
45.0
46.5
51.0
62.5
63.5
66.0
68.5
69.0
70.4
72.5
86.3
14.95
14.73
14.50
15.53
14.74
14.62
15.47
14.56
14.28
15.35
14.98
14.31
15.32
14.95
15.59
14.92
14.70
14.58
15.40
14.71
14.71
15.32
14.41
14.24
15.32
14.95
14.28
15.19
14.82
15.35
15.25
15.02
15.02
Pump Failed
15.63
14.91
14.68
Pump Failed
15.44
14.15
13.84
Pump Failed
14.53
14.23
13.18
Pump Failed
14.37
13.74
13.85
4.75
4.70
4.75
4.75
4.75
4.75
4.75
4.75
4.75
4.70
4.70
4.70
4.70
4.70
4.70
8.70
8.75
8.80
8.90
8.90
8.90
8.90
8.90
8.90
8.90
9.10
9.20
9.2
9.3
9.8
M.UUB. riux expresseo as gal/ftVday 8 77"F and 400 psi net. To con-
vert flux (gal/ft2/day 8 400 psi and 77° F) to Iiters/m2/day 8 2758
kN/m and 25° C, multiply by 40.67 and to convert psi to kN/m2, di-
vide by 0.145.
62
-------
Table 22. OPERATING PARAMETERS FOR TEST NUMBER TWO AT EBENSBURG, PENNSYLVANIA
en
Average recovery, percent 84.0 Avg.
Average feed rate, gpm
Average brine flow rate, gpm
Average recycle brine floiu rate, gpm
Average temperature, °F 63.3
Average feed pressure, psi
Average product flow rate, gpm
Normalized to 50°F
Normalized to 77°F
2
Average flux rate, gal/ft /day @ 50°F and 400 psi net
2
Average flux rate, gal/ft/day @ 77°F and 400 psi net
Maximum brine/product flow ratic
Minimum brine/product flow ratio
84.5 Max. 82.8 Min.
3.81
0.61
3.29
60.0 Min. 68.0 Max.
400
3.20
2.42
3.87
7.74
12.38
13.4:1
6.8:1
Dates of test : August 17-21, 1970
Length of run : 86.3 hours
Number of shutdowns : 4
Note: To convert flux (gal/ft2/day @ 400 psi and 77°F) to Iiters/m2/day § 2758 kN/m2 and 25°C,
multiply by 40.67; to convert gallons per minute to liters/second, multiply by 0.063; and to
2
convert psi to kN/m , divide by 0.145.
-------
Table 23. CHEMISTRY ANALYSES FOR TEST NUMBER TWO WHERE AVERAGE RECOVERY = 84.0%
en
PH
Conductivity
Acidity
Calcium as Ca
Magnesium as Mg
Aluminum
Total iron
Ferrous iron
Sulfates
Dissolved oxygen
All units are mg/1
(a)
Rpipr-hinn onualc
Raw Blended
feed feed
3.6 3.1
1470 4000
385 1160
210 650
63 200
37 110
150 460
98 280
1700 5200
2.0
Brine
3.0
6800
2150
1200
390
220
870
550
10,000
2.0
Product
4.6
85
58
1.8
1.1
1.1
1.9
<2
23
2.0
Re.iections(a)
97.9%
95.0%
99.7%
99.4%
99.0%
99.6%
99.3%
99.6%
^
except for conductivity (Mmhos/cm) and pH.
Blended feed concentration-product
concentration ., ,.
nn_
Blended feed concentration
-------
Test Number Three
Uithout brine recycle or acid injection, the RD unit would have
to treat pH 3.6 AMD at this site. It was the intent of Test IMumber
Three to simulate pH 3.6 influent conditions to the ^ K unit by in-
jecting sodium hydroxide to increase the pH of the blended feed.
The point of injection was highly critical in this application.
Referring to Figure ID, if the IMaOH were injected into the blended
feed, and overtreatment occurred, the ferric iron mould precipitate
directly in the RO unit. To protect the RO unit, the (MaOH was in-
jected into the raw feed prior to the sand filters. As the recycled
brine louered the blended feed pH from pH 3.6 to pH 3.1, the NaDH
increased the raw feed pH to counteract the effect of the recycled
brine. This meant the raw feed must be increased to approximately
pH 3.9 to obtain the desired result. At pH 3.9, however, the ferric
iron rapidly precipitated in the sand filter. The clogged sand fil-
ter prevented water flow to the unit so the test was discontinued.
The collected data were not sufficient to establish any definite trends.
Test IMumber Four
By eliminating brine recycle, the pH conditions which Test IMumber
Three uas trying to produce were achieved. With no recycled brine,
the raw AMD uas fed directly into the RD unit at pH 3.6. Unfortu-
nately, without the use of brine recycling, the small ^ H unit uas
limited to approximately 55 percent recovery.
Figure 1U shows the AP and flux history for the 3k hr, 50-55
percent recovery test. The AP remained constant throughout the
test run. Flux values, though someuhat erratic, uere identical for
tubes 1, 2, and 3. Flux history (Figure 1^+) did not indicate foul-
ing in any of the tubes. This was quite different from the readily
app3rent fouling seen in Tests IMumber One and Two. Peaks in the
flux value graph all occurred during periods of lowest water temper-
ature which indicate some discrepancies may exist betueen the manu-
facturer's temperature-flux normalization curve and the behavior of
the Ebensburg water.
65
-------
a.
° 7.0
Q
0)
3
0)
k
O.
o.
<
5.0
X
o
TJ
» a
^ O
~D °
O) "**
20-°
10.0
n-EHD-
-QO—a-
-Q
O TUBES 1&2
D TUBE 3
O
0
O
Q
i—
U
I
to
O TUBE 1
A TUBE 2
D TUBE 3
30
90
E
Z
40
30
D CS
~O e
800 « z
a> °°
«. f>
600
T3
C
u
Of o
400
60
ELAPSED TIME, hours
Figure 14.
Membrane performance andAP history for Ebensburg RO Test Number Four at
50 percent recovery
-------
Physical data for Test Number Four are presented in Table 24
and the chemical analyses are given in Table 25. Salt rejections
(Table 25) uere greater than 99 percent on all multivalent ions
except aluminum.
The downstream module in tube 3, which had been installed at
the beginning of Test Number Four, uas removed and a post mortem
analysis made. Sleigh reported that the module had been damaged
during shipment. He observed some iron present on the membrane
surface but uas unable to quantify it. There was no calcium sulfate
present on the membrane.
Discussion of Four Ebensburq Tests
At this site, it uas seen that pretreatment of the feed to louer
the pH uas not necessary for short-term tests as long as the pH uas
near 3.1 and/or the recovery uas near 50 percent.
Product uater quality, though uniformly good, uould require
slight addition of lime and filtering to be of potable quality.
As the fouling at 85 percent recovery could be removed success-
fully by a ueekly 3 hr lou-recdvery flush, fouling uould not pro-
hibit operation at that recovery level. The flux at 50 percent
recovery (Test Number Four), however, uas roughly 15 percent greater
than the flux during the 85 percent recovery runs (Tests Number One
and Tuo). Therefore, obvious improvements in membrane performance
can be obtained by operating belou the CaSO, fouling level.
Table 26 compares the results of the first Ebensburg test uith
^
those of REX Chainbelt at Mocanaqua. The spiral uound unit uas
operating at Ebensburg under more severe conditions in terms of
concentrations than the tubular unit at Mocanaqua, yet it produced
significantly superior results in terms of flux stability and product
quality. Due to the apparent success of the spiral unit at Ebensburg,
it uas not possible to determine if the fouling observed at Mocanaqua
uas peculiarly due to the tubular RO system or uhether some unknoun
characteristic of the Mocanaqua uater uas responsible.
67
-------
Table 2^. OPERATING PARAMETERS FOR TEST NUMBER FOUR AT EBENSBURG, PENNSYLVANIA
Average recovery, percent 53.2 Avg. 5*4.8 Max. 52.3 Min.
Average feed rate, gpm 6.33
Average brine flow rate, gpm 2.97
Average recycle brine flow rate» gpm 0
Average temperature, °F 60.2 Avg. 5^.0 Min. 63.0 Max.
Average feed pressure, psi ^01.1
Average product flow rate, gpm 3.36
Normalized to 50°F, gpm 2.78
g Normalized to 77°F, gpm k.k5
Average flux rate, gal/ft /day § 50°F and **00 psi net 8.90
Average flux rate, gal/ft2/day @ 77 °F and MDO psi net 1*«.2<*
Maximum brine/product flout ratio 8.7:1
Minimum brine/product flow ratio 5.5:1
Dates of test : August 31 - September U, 1970
Length of run : 93.5 hours
Number of shutdowns : 2
Note: To convert flux (gal/ft2/day @ UQO psi and 77°F) to Iiters/m2/day @ 2758 kN/m2 and 25°C,
multiply by U0.67; to convert gallons per minute to liters/second, multiply by 0.063; and to
2
convert psi to kN/m , divide by 0.1^5.
-------
Table 25. CHEMISTRY ANALYSES FOR TEST NUMBER FOUR WHERE AVERAGE RECOVERY = 53.2%
en
PH
Conductivity
Acidity
Calcium as Ca
Magnesium as Mg
Aluminum
Total iron
Ferrous iron
Sul fates
Dissolved oxygen
Ram feed
3.6
1160
390
160
51
30
130
96
1300
2.0
Brine
3.k
2160
770
330
110
61
250
210
2900
2.0
Product
<*.8
27
2k
1.1
0.3
1.1
O.i»
1.0
1.0
Rejections*
97.7%
93.956
99.3*
99.<»%
96.3%
99.5%
99.0%
99.9%
All units are mg/1 except for conductivity (Mmhos/cm) and pH.
-n . .. , Feed concentration-product concentration v lnn
*Rajection equals » . nnnnon+£U
-------
Table 26. COMPARISON OF MOCANAQUA AND EBENSBLJRG RO TEST RESULTS
Operating Parameters
Location
Investigator
Type of unit
Date of test
Length of test
Flux loss in 190 hours
Mocanaqua
Rex Chainbelt
Tubular
1969
813 hours
^40 percent (total unit)
Operating recovery (190 hours) :«70 percent
Ebensburg
EPA & Gulf Environmental Systems
Spiral-wound
1970
191 hours
zz25 percent (total unit)
:»84 percent
Chemical Parameters
Mocanaqua
Feed Product
pH
Acidity
Calcium
Sulfates
Total iron
Ferrous iron
Dissolved oxygen
Magnesium
3.6
—
140
790
100
100
4.9
100
4.1
—
2.8
36
3.7
3.7
—
1.9
Ebensburg
Feed Product
3.6
380
190
1640
135
HJO
2.0
54
4.1
56
1.2
14
1.7
>2
—
0.4
All units are mg/1 except for pH.
-------
Plans ijjere made to take the spiral RD unit to the Mocanaqua site
in the spring of 1971 to further investigate the situation.
NORTON FERRIC IRON LONG-TERM SPIRAL-WOUND STUDY
Following the Ebensburg study, the k K spiral unit was returned
to Norton for a long-term uinter study before going to Mocanagua in
the spring of 1971.
This Norton study lasted 3,013 hr at 73 percent recovery. The
flow arrangement (Figure 10) uas the same as previous studies except
that no pH control uas used. Neu high-flux modules uiith membrane
cast directly on the backing material were used for this test. One
of the major problems prior to this investigation uas the unreliabil-
ity of high pressure pumps to operate on AMD. A Gould MB 13600,
ceramic-stainless steel, multi-stage, centrifugal pump uas tested
on this study and performed flawlessly.
Flux history for the 3,013 hr test is shoun on a log-log basis
in Figure 15. Tube 3, uhose performance uas usually belou that of
tubes 1 or 2 due to its higher concentrations, exhibited superior
performance to tubes 1 and 2. Log-log slope changes occurred in
all tubes after approximately 300 hr. Until that time, tube 3 had
followed a log-log slope of -0.031 and tube one's slope uas -0.058.
Several flushing techniques, which are individually discussed
later, uere successful in restoring flux to values in excess of
those predicted by the initial slope. However, rapid flux losses
immediately recurred following each flush. The loss in tubes 1
and 2 was more severe than tube 3's loss. At the end of 3,013 hr,
tubes 1 and 2 had each lost 58 percent of their initial flux and
tube 3 had lost 38 percent.
The operating parameters for this study are given in Table 27
and chemistry analyses are presented in Table 28.
The data in Table 27 provide a possible explanation for the poor
performance of tubes one and two in respect to tube 3. Brine/prod-
uct flow ratios for this study were extremely low in general and
71
-------
X]
I\3
100
8O
70
6O
50
40
3O
o
c
S.20
o
o
•o
u.
o
£ 10
8
7
6
"X.
1 5
o
-o
-5 4
at
1
© TUBE THREE
• TUBE ONE
NOTE : 1 gal./ft2/day = 40.7 Iiter/m2/day
77°F = 25°C
600psi = 4138 kN/m2
I
I I I I ! I I
I
_L
I
10 2345 6789100 2 345 67891000 20003000
ELAPSED TIME, hours
Figure 15
Flux trends for Norton 3000 hour spiral-wound R.O. study at 13 percent recovery
-------
Table 27. OPERATING PARAMETERS FOR 3000 HR 4K NORTON STUDY AT 73
PERCENT RECOVERY
Parameter
Raw water feed flow
Product water flow
Brine water discharged
Brine water recycled
Minimum brine/product flow ratio
Maximum brine/product flow ratio
Water recovery
Recovery of blended feed
Feed pressure
Feed temperature
Tube one flux gal/ft/day @
2
Tube two flux gal/ft /day @
Tube three flux gal/ft2/day @
Unit
gpm
gpm
gpm
gpm
ratio/module
ratio/module
percent
percent
psig
op
77°F & 600 psi net
77°F & 600 psi net
77°F & 600 psi net
Value
5.00
3.6*+
1.36
1.46
2.8:1
6.6:1
72.8
56.3
600.2
54.4
17.43
17.25
20.41
Length of run
Date of run
hours 3012.6
November 16, 1970 - April 13, 1971
All values are means from 130 data sets.
Note: To convert flux (gal/ft2/day @ 600 psi and 77°F) to liters/
m2/day @ 4138 kN/m2 and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
73
-------
Table 28. NORTON UK 3000 HR CHEMICAL ANALYSES
Parameter
Raw feed
Blended feed
Brine
Product
Rejections
pH
2.9
2.7
2.5
3.9
Cond.
970
1600
2900
77
95.2%
Acidity
440
660
1600
22
96.7%
Cal Mg
120 39
180 71
420 140
1.1 .40
QQ /. Q£ QQ /.Q£
77«H^u 77*H7D
Fe
130
210
480
1.1
99.5%
Al
40
63
150
1.5
97.6%
*\
680
1100
2600
1.2
99.9%
„ . ., . Blended feed concentration-product concentration v inn
Rejection equals tiionn0H POOH mnnont^tinn X 10D
Blended feed concentration
All units are rog/1 except for conductance (micromhos/cm) and pH.
-------
at a minimum in tubes one and two (b/p minimum ratios of
2.8:1). The b/p ratio in tube 3 was significantly higher with
a maximum ratio of 6.6:1. It was felt that the lou brine/product
ratios did not provide sufficient turbulence in the brine stream
in the modules and thus did not effectively control boundary layer
precipitation.
Salt rejections (Table 28) mere uniformly above 99 percent on
all multivalent ions except aluminum.
Near the end of the test, copper sulfate uas injected directly
into the rauj feed water to determine the capability of the membrane
to reject copper. Concentrations of copper in the feed were varied
from 2,8 to ^ mg/1 resulting in blended feed concentrations from
4.*i to 60 mg/1. The membrane rejected an average of 99.5 percent
(based upon blended feed), yielding product copper levels below
D.16 mg/1 and brine copper concentrations up to 150 mg/1 at 75
percent recovery.
The severe fouling which occurred during this study afforded an
opportunity to investigate the effectiveness of various flushing
techniques. The flux history through 2,500 hr is shown on a linear
basis in Figure 16 to illustrate more clearly differences in flush
results.
The type of flush and its success in fouling removal should yield
some insight into the nature of the fouling. A list of the flushing
tried and their expected results follous:
1. Low-recovery Flush - Mainly scours the membranes due to
higher brine velocities and corresponding increases in
turbulence. Should remove some loose precipitation
such as flocculated ferric hydroxide or calcium sulfate.
May dislodge particulate matter. Mainly a physical
flushing technique.
2. Acidified Product Flush - Product water is stored and
acidified to pH 2.5 and then pumped through the RO unit.
75
-------
-o
en
»S
at
X
- 1000
15
NOTES
1 ACIDIFIED PRODUCT FLUSH AND
UNIT SHUT DOWN FOR ONE WEEK
DUE TO FLOODING
2 LOW RECOVERY FLUSH AND UNIT
SHUT DOWN OVER WEEKEND
500
1500
H800
u
o
CN
O
2
<
E
oo
CO
o
T3
CN
500
2000
1000
ELAPSED TIME, HOURS
Figure 16
Membrane performance during 4K reverse osmosis 3000-hr study at Norton
-------
This type flush should chemically redissolve those min-
erals uhich are acid soluble (i.e., iron, aluminum, cal-
cium, and some calcium sulfate) if the fouling is not
too severe before the flush is attempted.
3. BIZ flush - The catalytic action of enzymes should
mainly remove organics by breaking doun protein-contain-
ing materials. BIZ also contains uetting agents that
should aid in the resolubilization of precipitates and
phosphates that should be quite effective in sequestering
iron, calcium, and magnesium as well as in emulsifying
oils.
4. Shutdouns - Immediate depressurization should tend to
dislodge precipitates that have been held against the
membrane by pressure and RQ floiu through the membrane.
During long periods of depressurization, the membrane
tends to relax and the effects of compaction are reduced.
Also, normal osmotic flow (from product side into brine
side) takes place during this period and should tend to
dislodge precipitates from the brine side of the membrane.
Although all the flushes attempted (Figure 16) ujere successful
in improving flux, the most dramatic increase occurred using a tuo-
step flush. First, the unit uas flushed with acidified product
water, and then it was shut doun for one ueek because of flooding
uhich overloaded the filtration equipment. This accidental combina-
tion acid flushr-shutdOLjn at 725 hours elapsed time significantly im-
proved flux. Houjever, after each flush, flux again rapidly declined.
As each of the previously mentioned flushes were successful, the flux
decline uas attributed to a combination of compaction, iron fouling,
and organic deposition.
Samples were taken during the citric acid flush at 2,500 hours
elapsad time. Recovery during the flush uas maintained at 50 per-
cent so the brine concentrations should have been tuice that of the
rau feed. The resulting analyses indicated that the brine contained
77
-------
twice as much magnesium, aluminum, and sulfate as was in the acidi-
fied feed (normal far 5D percent recav/ery); however, eleven times
more iron and three times more calcium were in the brine than in the
feed. The predominance of iron in the flush brine indicated iron
precipitation was the major inorganic fouling mechanism.
A module was removed from tube 1 after 3,013 hr of operation at
Morton and dissected for membrane deposit analyses. An algal slime
covered the outside of the module. The membrane removed had an ob-
vious coating on it which, from its red-brown appearance, was tenta-
tively identified as ferric hydroxide.
For chemical analyses, the membrane was cut into small rectangles
(approximately 0.79 cm x 1.18 cm br 2 in x 3 in). Various reagents
were used to attempt to dissolve the coating from the membrane
patches. Table 29 lists the reagents used and chemical analyses of
the resulting solution. Initial analyses were made on each of the
reagents to determine background levels and to identify possible
interferences ujith the analytical procedures. Membrane patches ujere
weighed before and after immersion in the dissolving solution in an
attempt to correlate the resulting concentrations to a gravimetric
reference base (weight of coating). In most cases, however, the
solutions dissolved the membrane as well as the coating and only the
fabric backing material remained; therefore, the gravimetric refer-
ence base was ineffective as it was not possible to ascertain the
amount of membrane removed when only partial membrane destruction
occurred. The membrane coating appeared to consist mainly of iron
and-sulfate (Table 29). Presence of sulfate without calcium was not
completely understood.
Samples of the membrane were sent to Rex Chainbelt and Gulf
Environmental Systems for analysis. Mason of Rex Chainbelt found
two distinct layers of fouling material on the membrane:
"The first layer closest to the membrane was an iron
oxide coating. The second layer was a clay-like sub-
stance which was approximately 65 percent organic ma-
78
-------
Table 29. CHEMISTRY ANALYSES OF COATING MATERIAL ON 4K REVERSE OSMOSIS MEMBRANE
UD
Sample
designation
Blank
Stannous chloride
Blank
Sodium hydrosulfite
Blank
Citric acid
Blank
Hydrochloric acid
Blank
Phosphoric acid ,, ^
Stannous chloride,. J
Stannous chloride /•. •»
Sodium hydrosulfite^ x
Sodium hydrosulfite
Citric acid
-------
terial. This second layer adhered to the mem-
brane very loosely, and was easily removed hy
slight egitation. The ease of removal may have
been due to the handling of the membrane when
removed from the pressure vessel and shipped
to Milwaukee. The membrane flux, tested with
both layers of fouling material, was 12.7
gel/ft^ dav at U15 psi and 77° F. Salt rejec-
tion was approximately 96 percent sodium chlo-
ride. The membrane was flushed with R 2 mt.
percent of sodium hydrosulfite (IMa^S,-^ ) in
water. The pH of this solution was approxi-
mately 6.0. This chemical solution provided
a strong reducing environment which converted
the ferric iron to soluble ferrous iron, and
thus allowed cleaning of iron oxides from the
membrane. This cleaning can generally be effected
with a two-hour flush. After cleaning by the
above procedure, the membrane had a flux of
16.6 gal/ft? day at £*15 psi and 77" F. The salt
rejection remained unchanged at 96 percent, and
therefore the membrane surface was not damaged.
It should be noted that the second layer of
fouling material was not dissolved by the sodium
hydrosulfite solution, and therefore must be re-
moved by other methods, such as normal osmosis
flushing, enzyme solutions, and high velocity
flushing. "
1^
Sleigh reported the analysis of the coating material to be fer-
ric hydroxide and mud.
Since the product mater was not of potable quality, a feasibility
test was made for lime neutralization to increase pH and remove resid-
ual iron.
80
-------
During the test, product from the RD unit uias diverted to a
200 gal reaction tank equipped with a mixer. A pH controller auto-
matically controlled the addition of a 1 percent slurry of lime to
maintain a pH between 7.4 and 7.8 in the reaction chamber- Neutral-
ized product water mas then pumped through a 10 micron cartridge
filter.
Table 30 shows the effectiveness of product neutralization.
Table 30. CHEMISTRY ANALYSES FOR RO PRODUCT NEUTRALIZATION
Acid-
Uater pH Alk. Cond. ity Ca Mg Fe Al SO,
Product 3.7
Neutral
product 7.4
0
5
55
27
9.3
0
0.37 0.20 0.25 0.19 0.33
5.1 0.20 0.04 0.10 0.40
All units expressed as mg/1 except conductance (Mmhos/cm) and pH.
An increase in calcium was expected since lime [Ca(OH) J uias
used as the neutralizing agent. With the decrease in iron and acid-
ity, the neutralized product was of potable quality.
Continuous RO product neutralization required (9.83 gram/m )
0.082 pounds of lime per 1,000 gal.
Discussion erf Norton Study
Fouling, which was observed during this long-term study, was severe
toward the end of the run. The fouling mechanism uas diagnosed as a
combination of organic and colloidal deposition and ferric hydroxide
precipitation. Several flushing techniques were successful in short-
term flux restoration, but rapid degradation recurred after each
flush.
The apparent reason for the fouling uas operation of the spiral
system at insufficient brine flou rates thus allowing boundary layer
81
-------
precipitation. This conclusion uas confirmed by superior flux sta-
bility of tube 3 as compared to tubes 1 and 2. Tube 3 performance
normally ujould be inferior since it uas subject to the most severe
concentration of pollutants. In this study, however, the superiority
of tube 3 uas attributed to a brine/product flow ratio which uas up
to 2.^ times the b/p ratio in tubes 1 and 2.
Injection of copper sulfate into the RO feed water determined
that the membrane uould reject copper at a 99.5 percent rejection
rate.
RD product water uas not of potable quality due to iron and pH.
Lime neutralization of the product uater resulted in a potable qual-
ity effluent.
Further testing uas necessary at higher brine/product flow ratios
to provide more representative long-term flux trends. Additional
tests along this line were made after the <* h unit returned from
Mocanaqua in Fall 1971.
MQCAIMAQUA FERROUS IRON STUDIES COMPARING SPIRAL-UOUND, HOLLQLJ-FIBER,
AIMD TUBULAR UNITS (15,16)
To investigate the Mocanaqua fouling phenomena, EPA and the
Commonuealth of Pennsylvania contracted uith Rex Chainbelt in 1971
to evaluate the iron fouling problem in the laboratory, to modify
the 1969 tubular system, and to conduct additional studies at
Mocanaqua. In order to make these studies more comprehensive, EPA
was to conduct simultaneous studies uith the k K spiral-uound RD
unit at the same site. A hollou-fiber RO permeator uas also
obtained for this study, and EPA later enlarged it to a three perm-
eator array for a hollou-fiber system of comparable size to the
spiral-uound system.
For the ensuing study, all three systems operated side by side
on the same uater; the only exception uas the acid injection to the
feed during the last phase of spiral-uound testing. Thus, a unique
opportunity uas provided for direct comparison of systems and for
investigation of the Mocanaqua iron fouling phenomenon.
82
-------
Typical uater quality characteristics of the Mocanaqua discharge
are presented in Table 31.
Table 31. TYPICAL RAU UATER QUALITY CHARACTERISTICS
DP MOCAWAQUA DISHCARGE
Parameter
pH
Conductance
Acidity
Calcium
Magnesium
Total iron
Ferrous iron
Aluminum
Sulfate
Manganese
Silica
TDS
Dissolved uxygen
Temperature
Units
Mmhos/cm
mg/1 as CaCO-,
mg/1
ii
n
M
n
ii
n
ti
n
n
°Fahrenheit
Value
3.<+
1100
230
120
90
80
68
11
800
15
10
1200
1
5k
The Rex Chainbelt Laboratory, in addition to the normal EPA
determinations, analyzed for silica by atomic absorption. A Hach
Kit uas used by Rex Chainbelt for on-site determinations of calcium,
total hardness, total iron, ferrous iron, and sulfates. A Myron L
TDS meter uas used for total dissolved solids measurements.
As a prelude to actual field investigations, laboratory studies
were performed by Rex Chainbelt to attempt to isolate possible causes
for the severe iron fouling seen at Mocanaqua during their 1969 tests.
From the results of these batch tests, Mason postulated that the
most probable cause of the observed fouling uas bacterial oxidation
of ferrous iron at the membrane surface and subsequent precipitation
of the resulting ferric iron uhich had been hydrolyzed.
For this reason, ultraviolet (UU) disinfection lights uere in-
stalled as part of the pretreatment system for all three field units
tested in the belief that the U\J light would kill most of the iron
oxidizing bacteria and thus inhibit or prevent bacterial oxidation
9 3
of Fe to Fe .
83
-------
Iron oxidation studies ujere performed during the actual field
study to evaluate the effectiveness of the UV light. Sixteen sets
of samples uere taken for one of these tests. Twelve of the 16
were raw feed samples that had passed through ID micron filters.
The remaining four sets were blended feed samples from the spiral
RD unit. Each individual set of samples consisted of five dupli-
cates. Ferrous iron was determined once per day on each set of
samples for a period of five days. To prevent contamination from
the pH probes and stirrers which uere used to determine ferrous
concentrations, only one of the duplicates was analyzed per day.
Following analysis, the sample uias considered contaminated and was
discarded.
All samples uere kept in a mine discharge at a 11°C (53°-54°F)
temperature to eliminate environmental changes. Those samples to
be kept dark uere placed in opaque plastic bags, placed in the
creek, and covered with brush to provide shade. Sulfuric acid uas
added to those samples in which the pH was lowered. Formaldehyde
uas used for a disinfectant uhere required. The results of the
test are given in Table 32. Specific conclusions relative to passi-
ble oxidation control techniques from the oxidation study were:
• ultraviolet disinfection inhibited oxidation;
• lowering the pH to 2.5 inhibited oxidation;
« some oxidation occurred in blended feed samples which
had been exposed to UU and kept in dark;
• at creek temperature (5k° F), virtually no oxidation
took place in 2k hours; and
• the oxidation mechanism uas bacteria since the above
variations in treatment inhibited oxidation.
Since some oxidation occurred in two of the blended feed samples
that had been exposed to ultraviolet treatment, it uas probable that
the U\7 did not effect a 100 percent kill. If bacteria were present
in the blended feed, it uas assumed that significant growth uas
probable at the membrane surface in the RO units.
-------
Table 32. FERROUS IRON OXIDATION CONTROL STUDY, MOCANAQUA, PA.
CD
LD
Ferrous concentration after:
Sample
Number
1
2
3
k
5
6
7
8
9
10
11
12
13
1<*
15
16
Note:
(a)n-n
2k
Initial hours
56
56
56
56
56
56
56
56
56
56
56
56
118
118
118
118
Results
56
56
56
56
56
56
56
56
56
56
56
118
112
118
118
expressed as
hours
56
56
56
56
56
56
56
39
56
39
56
118
112
112
118
mg/1.
72
hours
56
22
56
56
56
56
56
56
28
56
56
118
79
96
118
-.,-, fi 1 •*.„,
96
hours
56
1.8
56
56
56
56
56
56
28
56
28
56
118
39
96
118
Description(a'c)
Raw feed - light
Raid fpprl — rlark
Rau, ultraviolet-light
Rau, ultraviolet-dark
Raw, pH 2.5 - light
Raw, pH 2.5 - dark
Rau, UU, pH 2.5 - light
Raw, UU, pH 2.5 - dark
Raw, UU, disinfecttb; - light
Rau, UU, disinfect(b) - dark
Raw, disinfectib; - light
Rau, disinfect - dark
Blended feed, UU - light
Blended feed, UU - dark
Blended, disinfectCb\ UU -
Blended, disinfect(b), UU -
light
dark
\ ^Disinfection consisted of addition of 2 ml of formaldehyde to <*00 ml/sample.
All temperatures 53-5**°F except for initial blended feed sample temperature of 6**° F.
-------
Another series of oxidation tests at air temperatures 15.6° to
23.9° C (68-75° F) resulted in significantly faster oxidation rates
as is typical in most biological growth.
TUJO sets of spiral-wound modules were used in the 4 K unit
2
during the study. During Phase I, each module contained 4.65 m
(50 ft2) of standard percholate-modified cellulose acetate membrane
2 2
for a total unit membrane area of 41.9 m (450 ft ). The modules
2 2
used for Phase II contained 5.77 m (62 ft ) of Formamid-modified
2 2
cellulose acetate for a total unit membrane area of 51.9 m (558 ft ).
Figure 17 presents the system arrangement and flow diagram for
the spiral unit. Acid mine water was pumped through ten micron car-
tridge filters and a LJU disinfection light before entering the re.-
vsrse osmosis unit. UW disinfection was believed necessary to
inhibit bacteriological oxidation of the ferrous iron uhich had
ii
presumably contributed to fouling observed in earlier studies.
During Phase II, sulfuric acid was injected to lower the blended
feed pH to 2.9 and thus inhibit both ferrous oxidatidn and ferric
precipitation.
The second unit tested was a hollow-fiber system, manufactured
by DuPont; initially, it consisted of one permeator on loan from
DuPont to Rex Chainbelt. As illustrated in Figure 3, approximately
2 2
139.6 m (1500 ft ) of B-9 hollou-fiber nylon membrane were packed
in a 15.2 cm x 1.22 m (6 in x 4 ft) stainless steel pressure vessel.
During Phase I, the single DuPont permeator operated at 2758 k(\l/m2
(400 psi) from a Moyno high pressure pump which also supplied feed
water to the tubular unit. Phase I DuPont tests were successful
so EPA purchased two additional permeators for Phase II and formed
a three permeator array, 6 K unit, as shown in Figure 18. The brine
from both new psrmeators in parallel served as the feed to permeator
number 3 (the original Phase I permeator). Flow control orifices
were placed in each brine line from permeators 1 and 2 to ensure
equal brine flows and prevent system unbalance. Because high brine
flows are not required for hollow fiber systems, it was not necessary
86
-------
CD
-O
Blended
Raw
Feed
Feed
Tube 1
Tube 2
Tube 3
Recycled Brine
1892
Liter
Supply
Tank
High Pressure
Pump
Deep Mine Portal
r—I.jn (50° sal)
j Acid J
(Injection)
I I
Product Water
Waste
Brine
Figure 17. Spiral -wound reverse osmosis system arrangement at Mocanaqua, Pa.
-------
Raw Feed
33
03
Module 1
Flow Control
Orifice
Module2
Module 3
Brine
Product Water
Ultra-
Violet
Light
High Pressure Pump
Deep Mine Portal
Figure 18. Hollow-fiber reverse osmosis system
arrangement at Mocanaqua, Pa.
-------
to recycle brine. Therefore, the unit received raw acid mine water
that had been filtered and had passed through the ultraviolet light
for disinfection.
Phase I tubular studies ' utilized a unit consisting of sixty
7.6 cm x 2.39 m (3 in x 7 ft ID in) plastic tubular #310 modules
illustrated in Figure 19 and arranged as shown in Figure 20. The
tubular system was manufactured by Calgon-Havens Company. Each
module contained eighteen 1/2 in (1.27 cm) i.d. porous fiberglass
tubes that were lined with cellulose acetate membrane and connected
2 2
in series. Membrane area was 1.57 m (16.9 ft ) per module for a
2 2
total system membrane area of Vk.k m (101^ ft ). Pretreatment was
the same as that for Phase I Gulf and DuPont, i.e., 10 micron fil-
tration and UU disinfection.
Turbulence promotion rods were placed in modules where brine
flow was lowest to inhibit 'boundary layer1 precipitation and fouling.
Unfortunately, use of turbulence promoters also resulted in high
pressure drops across the modules in which they were installed, thus
reducing the average applied pressure.
During Phase II, the sixty #310 modules were replaced with five
2 2
#610 modules for a total membrane area of 7.87 m (8*4.5 ft ).
It was necessary to determine the osmotic pressure of the water
in order to compensate for variations in water quality and recovery,
Ik
The spiral-wound unit was modified to enable measurement of osmotic
pressure by pressure differences between the brine and product side
of the membrane. By varying recovery, osmotic pressure measurements
were made over a complete range of concentrations and a correlation
was developed between brine conductivity and osmotic pressure.
Since acid injection was utilized in Phase II spiral-wound studies
and since acid affects conductivity, a different correlation was
necessary for Phase II. Both equations and plots are shown in
Figure 21.
89
-------
SINGLE TUBE WITH TURBULENCE PROMOTER ROD
TURBULENCE
PROMOTER
ROD END VIEW
18 POROUS
FIBER GLASS TUBES
IN SERIES
PRODUCT
WATER SHROUD
FEED WATER
MODULE ASSEMBLY
PRODUCT WATER
Figure 19
Tubular reverse osmosis module configuration' ' at Mocanaqua, Pa
-------
Raw Feed
,\\
H I
H 1
i i i
h
i
i i i h
-n i i i \\-
-D
l l I
-T
H
n_
_
1 1 I
1 1 1
1
Ih
H
L
EachQ represents one module as shown in Figure 19
r| 1
^ i
-i 1 1
M 1 1
\.
*" \Filter / —
Ultra
1 Ink*
Wh
1
lV|Yl|-
1
|V|VH-
I
|v VJ|J
1
H 1 IVIY
EH
H 1 IYIYIY
»• H
igh Pressure Pump
••!•••
l| Waste Brine
J
Product Water
1
H8>-J
60 modules @ 1.57m* (16.9 ft* ) membrane each 3 banks,
5 modules in series, each row modules marked with a V have
volume displacement (turbulence promoter) rods
Deep Mine
Portal
Figure 20* Phase I module arrangement
study at Mocanaqua, Pa.
06)
for tubular reverse osmosis
-------
- 200
- 150
in
a.
VD
U)
4)
O
E
Ifl
O
No Acid
Injection
100
v)
VI
4)
ito
a.
u
«•
O
E
v>
O
1000 2000 3000 4000
Conductivity (C) micromhos/cm
Figure 21
Osmotic pressure - conductivity relationships at Mocanaqua, Pa.
-------
Spiral-Wound Phase I Studies
Phase I of the spiral-wound studies was made at a unit recovery
O
of 75 percent and system operating pressure of 4138 kl\l/m (600 psi).
A summary of operating parameters for this study is given in Table 33.
Water temperatures of 63° F were caused by the Gould high-pressure
pump that imparted energy to the 54" F ram AMD. The necessity of
brine recycle only aggravated the temperature rise since a portion
of the uater was continually recycled through the pumps. The in-
creased temperature benefitted observable flux rates since product
flDu uas strongly temperature dependent. Since all data were cor-
rected for temperature, the higher operating temperature uas not
a factor in data evaluation. In actual operation, however, the
V
higher the temperature, the higher the flux. Overall system recov-
ery during Phase I uas maintained at 75 percent during the first
1,054 hr of operation. The log-log flux decline slope uas -0.034
for tube 1 and -0.032 for tube 3 during this period. At 1,054 hr
elapsed time, the ultraviolet light uas turned off to determine its
effectiveness in controlling flux decline. As shown in Figure 22
(linear plot) and Figure 23 (log-log plot), the increase in flux
decline was dramatic with the log-log slope increasing to near
-0.2 even though recovery remained at 75 percent. Following a
phosphoric acid flush, the U\J uas turned on again but recovery uas
increased to 84 percent. Flux decline uas so severe at 84 percent
recovery (log-log slope greater than -4.0) that the run was termin-
ated after only 70 hours at that recovery level and flushed again
uith phosphoric acid. Though the acid flush improved flux, it did
not stabilize on the original slope, presumably indicating all the
fouling had not been removed. Total operating time during Phase I
was 1,922 hr.
Fouling observed at 84 percent recovery was attributed to cal-
cium sulfate precipitation. Analysis of the acid flush water con-
firmed the presence of large amounts of both iron and calcium after
the 84 percent recovery run. A loss of calcium, sulfate, and iron
93
-------
Table 33. OPERATING PARAMETERS FOR SPIRAL-WOUND PHASE I STUDY AT
75 PERCENT RECOVERY AT MOCANAQUA, PENNSYLVANIA
Parameter Value
Raw water feed flow, gpm 6.02
Product water flow, gpm 4.50
Brine water discharged, gpm 1.52
Brine water recycled, gpm 4.26
Minimum brine/product flow ratio (Tubes 1 and 2),
ratio/module 5:1
Maximum brine/product flow ratio (Tube 3),
ratj.o/module 12:1
Water recovery, percent 74.8
Recovery of blended feed, percent 43.8
Feed pressure, psig 602
Feed water temperature, °F 62.6
Tube one flux, gal/ft2/day @ 600 psi net and 77° F 19.56
Tube two flux, gal/ft2/day @ 600 psi net and 77° F 19.52
Tube three flux, gal/ft2/day @ 600 psi net and 77° F 18.77
Length of run, hours 1672
Date of run May 1 - July 12, 1971
All values are means from 73 data sets.
Note: To convert flux (gal/ft2/day @ 600 psi and 77°F) to liters/
m /day 8 4138 kN/m2 and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
-------
85% Recovery
Ul
O Tube 1
• Tube 3
o
ui
CN
TJ
C
O
CM
1200
1000
00
CO
o
TJ
CN
800
600
4-
4-
1-
300
600
900
1200
1500
1800
2100
ELAPSED OPERATING TIME, hours
Figure 22
Phase I spiral-wound reverse osmosis unit flux rates at
Mocanaqua, Pa.
-------
UD
cn
a
TJ
o
c
-o
c
a
o
30
20
O
O)
at
X
10
8
7
6
5
4
3
Note: Recovery 75% Except Where Noted
Spiral Wound RO
Unit Phase I
Tube 3
Tube 1
Spiral Wound RO Unit Phase II 400PSI
Tube3
Q/
00
Slope-0.01
Hollow Fiber RO Unit
Phase II 400 PSt
o>
Note: Module 3 Total Elapsed Time is +838 Mrs
From Phase I
1 gal/f ft2/day=40.7 Iiter/m2/day
77°F = 25°C 600 psi= 4138 kN/m2 400 psi- 2759 kN/m 2
I I I I I I I I
10
20 30 50 100 200
ELAPSED TIME, hours
Figure23
500
1000
2000
A summary of reverse osmosis flux trends at Mocanaqua, Pa.
-------
was detected in the routine chemical analyses as noted in Table 35.
At 75 percent recovery and the U\J off, it was felt that bacterial-
propogated iron precipitation uas causing the flux decline.
Operating parameters for the spiral-wound unit during the 84
percent recovery run are given in Table 34.
Chemical analyses for the 75 percent recovery operation (Table
35) indicate a 99+ percent rejection on all multivalent ions. Prod-
uct waters would still require treatment for iron and manganese and
pH adjustment before potable water quality criteria could be met.
Spiral-Mound Phase II Studies
The manufacturer suggested that increasing the b/p flow ratio
would decrease the log-log slope observed during Phase I. One way
of accomplishing a b/p increase was to lower the operating pressure
and thereby reduce product flow. Lowering input pressure to
2
2758 kIM/m (400 psi) would also serve to lessen membrane compaction
which contributes to flux losses. During Phase I, the maximum b/p
ratio per module was 12:1 and the minimum was 5:1 (from Table 33).
During Phase II, the maximum b/p ratio was 22:1 and the minimum
10:1 (Table 36) which amounted to about a 200 percent increase over
Phase I.
As the earlier oxidation tests had shown pH control to be an
effective oxidation inhibitor and as the lower pH would also inhibit
precipitation of the ferric iron already present, sulfuric acid was
injected into the feed water to lower the blended feed pH to 2.9.
The UV light was also used during the majority of this Phase II
study.
Table 36 presents the operating parameters for this 2,454 hr
Phase II study at 75 percent recovery.
Early in the study, leaks were observed in tubes 1 and 2. By
probing each module and measuring the conductivity of product water
at various points along the inside of the modules' product tubes,
97
-------
Table 34. OPERATING PARAMETERS FOR SPIRAL-WOUND PHASE I STUDY AT
84 PERCENT RECOVERY AT MOCANAQUA, PENNSYLVANIA
Parameter
Raw water feed flow, gpm
Product water flow, gpm
Brine water discharged, gpm
Brine water recycled, gpm
Minimum brine/product flow ratio (Tubes 1 & 2),
Value
4.92
4.13
0.79
5.50
ratio/module 8:1
Maximum brine/product flaw ratio (Tube 3),
ratio/module 12:1
Water recovery, percent 83.9
Recovery of blended feed, percent 39.6
Feed pressure, psig 600.0
Feed water temperature, °F 66.6
Tube one flux, gal/ft2/day S 600 psi net and 77°F 17.03
2
Tube two flux, gal/ft /day @ 600 psi net and 77°F 17.06
Tube three flux, gal/ft2/day @ 600 psi net and 77°F 16.25
Length of run, hours 70
Date of run July 13 - July 17, 1971
All values are means from four data sets.
Note: To convert flux (gal/ft2/day @ 600 psi and 77°F) to liters/
2 2
m /day @ 4138 kN/m and 25°F, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
98
-------
Table 35. CHEMICAL ANALYSES FOR SPIRAL-WOUND STUDIES AT MOCANAQUA, PENNSYLVANIA
Sample
Designation
pH
Cond.
Acid-
ity
Cal-
cium
Spiral-mound
Rau feed
Blended feed
Brine
Product x ,.
Rejections a
3.4
3.1
2.9
4.4
1080
2070
3540
17
99.2%
240
460
810
38
91.7%
130
260
490
0.4
99.8%
Spiral-wound
Raw feed
Blended feed
Brine(b)
Brine
Product
3.4
3.0
2.9
4.4
1010
2800
4800
23
210
700
(1260)
1290
10
180
400
(1080)
760
0.7
Spiral-wound
Rau feed
Acidified feed
Blended feed
Brine
Total product
Overall re-
jections'8)
Tube 1
Rejections'8^
Tube 3
Rejections^8)
3.4
2.9
2.8
2.7
3.8
1110
1520
2830
4150
67
97.6%
86
97.0%
37
98.7%
220
420
610
880
33
94.6%
_
140
140
330
520
1.3
99.6%
1.6
99.5%
0.55
99.8%
Magne-
sium
Total
iron
Phase I @ 75 Percent
88
170
310
0.3
99.8%
77
180
330
0.4
99.8%
Phase I @ 85 Percent
130
380
(780)
710
0.8
Phase II @
110
110
260
380
4.0
98.5%
3.1
98.8%
0.75
99.7%
140
310
(840)
540
0.5
Ferrous
iron
Alum-
inum
Sul-
fate
Man- Dissolved
qanese oxyqen
Recovery
64
130
250
0.3
99.8%
12
24
44
0.2
99.2%
750
1340
2300
0.9
99.9%
- < 1.0
<1.0
<1.0
Recovery
78
250
(470)
440
0.35
15
55
(90)
100
0.5
850
1420
(5100)
2840
1.0
75 Percent Recovery
100
100
250
370
2.4
99.0%
3.3
98.7%
0.68
99.7%
73
77
190
280
2.0
98.9%
_
-
_
14
14
33
54
0.8
97.6%
1.1
96.7%
0.75
97.7%
930
980
2110
3130
19
99.1%
30
98.6%
5.8
99.7%
17
17
43
68
0.5
98.8%
0.6
98.6%
0.21
99.5%
^Rejections = (blended feed - product)/blended feed X 100.
Brine values in parenthesis mere calculated based on the recovery of the reverse osmosis unit and assuming
the raw feed values mere correct. At 83.4 percent recovery, the concentration factor is 6.02. This calcu-
lation was necessary since massive precipitation occurred in the brine sample before analysis could be made
at Norton.
Note: All units are mg/1 except pH and specific conductance (micromhos/cm).
-------
Table 36. OPERATING PARAMETERS FOR SPIRAL-WOUND PHASE II STUDY AT
75 PERCENT RECOVERY AT MOCANAQUA, PENNSYLVANIA
Parameter ___________ Value
Rau water feed flow, gpm 5.IB
Product water flow, gpm 3.86
Brine water discharged, gpm 1.32
Brine water recycled, gpm 6.32
Minimum brine/product flow ratio (Tubes 1 & 2),
ratio/module 10:1
Maximum brine/product flow ratio (Tube 3), ratio/module 22:1
Water recovery, percent 74.5
Recovery of blended feed, percent 33.6
Feed pressure, psig 399
Feed water temperature, °F 66.6
Tube one flux, gal/ft2/day @ 400 psi net and 77°F 12.81
Tube thio flux, gal/ft2/day § 400 psi net and 77°F 12.55
Tube three flux, gal/ft2/day @ 400 pai net and 77°F 11.58
Length of run, hours 2,454
Date of run July 27 - November 9, 1971
All values are means from 100 data sets.
Note: To convert flux (gal/ft2/day @ 400 psi and 77°F) to liters/
2 2
m /day @ 2758 kN/m and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
100
-------
all leaks were found to be near the ends of the modules where glue
connected the membrane to the perforated product tube (see Figure 1).
The modules weren't replaced because considerable operating time
would be lost and since modules in tube 3 had no leaks, their product
quality would be representative of normal modules. Also, as the
leaks in tube 1 and 2 were very small, they would have no effect on
flux studies. Therefore, chemistry data in Table 35 are given for
product water from tubes 1 and 3 as well as for the entire unit.
Although unit rejections were lower for Phase II than for Phase
I because of the leaks, tube 3 rejections were comparable to Phase
I results.
A log-log plot of Phase II flux is given in Figure 23. Log-log
slopes of -0.012 for tube 1 and -0.009 for tube 3 were obtained
during the 2,^5^ hr of operation at 75 percent recovery. These
values were improvements over the -0.03^ slopes from Phase I. The
combination of acid injection to inhibit precipitation of ferric
hydroxide and the higher b/p ratios and brine flow rates were be-
lieved responsible for the flux stabili?ation. Discontinuing use
of the ultraviolet light at 2,002.5 hr elapsed time had no apparent
effect on flux stability.
Possibly the higher b/p ratio and correspondingly higher brine
flow rates—or acid injection alone—may have been solely responsible
for flux improvement, but unfortunately, no tests were made to prove
or disprove this conjecture.
A phosphoric acid flush at the end of the 2,^54 hr study did
not improve flux.
Approximately 60 ml of sulfuric acid (98 percent, technical
grade) was required to treat each 1,000 gal of water entering the
RO unit in order to lower blended feed pH to 2.9. This amounted to
27 mg of concentrated acid per liter of water treated.
101
-------
Hollow-Fiber Phase I Studies
Phase I hollow-fiber testing began with a single permeatar
2
operating at 75 percent recovery at 2758 k!\!/m (400 psi) and con-
tinued for a total of 838 hr- Operating parameters are given in
Table 37 and chemical data are presented in Table 40.
Rejections of all multivalent ions were in the range of 99 per-
cent based on raw feed. These rejections, though slightly less than
the spiral-ujound, uere still excellent. As ujith the spiral unit,
the product water would require treatment for pH, iron, and manganese
before potable standards could be met.
The flux trend for the single hollow-fiber permeator is shown
linearly in Figure 24 and on a log-log basis in Figure 25. The log-
log flux decline slope was O.D37 during the 838 hr study. This flux
decline slope was comparable to Phase I spiral-wound results. A
small increase in pressure drop across the permeator was observed
which probably indicated minor iron fouling.
PhaseII Hollow-Fiber Studies
For Phase II studies, two additional permeators were added to
the Phase I system to form a 2-1 array as shown in Figure 17. The
Phase I permeator, which had accumulated 838 hr of operation, was
used as the last permeator in the array. The permeators were opera-
ted in this 2-1 array to allow high recovery experiments and confine
the expected calcium sulfate fouling to a single permeator. Gener-
ally, hollou)-fiber units are not used in staged arrays but are
arrayed for parallel operation since brine flow requirements for
hollow-fiber systems are not as critical as for tubular and spiral-
wound systems.
During Phase II, several variations in operating parameters were
made. In order to evaluate the effect of the ultraviolet light on
system operation, the light was turned off from 140 hr elapsed time
(e.t.) through 595 hr e.t. This time interval represented 978-1433
hr e.t. on permeator number 3 (last stage permeator). As seen in
102
-------
Table 37. OPERATING PARAMETERS FOR HOLLOW-FIBER PHASE I STUDY AT
75 PERCENT RECOVERY AT MOCANAQUA, PENNSYLVANIA
Parameter
Raid water feed flow, gpm
Product mater flow, gpm
Brine water discharged, gpm
lijater recovery, percent
Feed pressure, psig
Feed water temperature, °F
Unit flux, gal/ft2/day @ 77°F and 400 psi net
Value
2.08
1.55
0.53
7k. 4
400.0
53.3
2.32
Length of run, hours 638
All values are means from 78 data sets.
Note: To convert flux (gal/ft2/day @ 400 psi and 77°F) to liters/
m2/day @ 2758 kN/m2 and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
103
-------
00
O
-o
c
o
o
o
E
Q.
O>
t—
u
13
Q
O
of
0.
2.25 -
2.0 -
1.75 —
Operated as a Single Permeator
Water Recovery - 74.3 ± 0.28%
I— Brine Flow - 0.54 ±. 0.01 gpm
0.14
200
400
600
800
2.0
1.75
1.5
Operated as Final Stage
Permeator in a 2-1 Array
I Overall Water Recovery
Percent
I
I
0.12
0.11
0.10
0.09
I
II
III
IV
76.2 ± 0.6
84.4 ± 0.7
78.5± 0.5
75.3 ± 0.5
800 1200 1600 2000 2400
ELAPSED TIME, hours
2800
Figure 24
Operational history of hollow-fiber permeator #691
during the Mocanaqua study
o
CN
-o
C
o
CN
E
10
K
CN
u
D
a
O
Qi
a.
-------
a
ui
I III
Tubular RO Unit Phase 2 600 PSI
Slope-0.04
Slope-0.037
Note:l gal/ft2/day=40.7 Iiter/m2/day
77°F = 25°C Hollow Fiber Permeator
,si = 4138kN/m2 #691(400PSI)
I
See Figure 24 for
Details of Run
I I
10
20
40 70 100 200 400
ELAPSED TIME, hours
700 1000
2000 3000
Figu re 2 5
Tubular and hollow-fiber reverse osmosis unit flux
trends during the Mocanaqua study.
-------
Figure 23, no apparent flux trend change occurred as a result of no
UU. This was opposite to the dramatic slope change which occurred
on spiral-uound Linen the UU was turned off.
At 390 hr through 426 hr, e.t., the system recovery was increased
to 85 percent. This increase in system recovery had very little
effect on the recoveries of permeators 1 and 2 whose recoveries were
still below 65 percent. Thus, very little change in flux occurred
in the first two permeators during this high system recovery opera-
tion. Permeator 3, however, which was subject to the most severe
pollutant concentrations, suffered a drastic flux loss. Upon lowering
recovery again to 75 percent, permeator 3's flux again stabilized
(Figure 23).
At 596 hr elapsed time (1,434 for permeator 3), the UU light was
turned on again but system recovery was increased to 85 percent.
These operating conditions were maintained for 111 hr, including
operating at SO percent recovery for 38 hr, before system recovery
was lowered (707 hr e.t., 1,545 e.t. for permeator number 3).
Again, severe flux losses occurred in permeator 3 (log-log slope
= -1.77). Although the flux in permeator 3 stabilized, it was now
following a significantly steeper flux decline slope (log-log slope =
-0.36 at 78-80 percent recovery). A subsequent flush with disinfect-
ant failed to improve flux. Sodium hydrosulfide was used to flush
permeator 3 at 2,081 hr (on permeator 3) and successfully restored
flux to the original flux decline slope of -0.037 at 78 percent re-
covery. Flux immediately dropped again, however, but stabilized at
lower values until the end of the run at 2,670 hr e.t. (permeator 3).
Permeators 1 and 2 appeared to suffer a flux slope change near
600 hr elapsed time after following an initial slope of -0.015 until
that time. Following the 85 percent recovery test which ended at
707 hr elapsed time, first stage permeator flux stabilized following
a log-log slope of -0.03 for approximately 540 hours until 1,243 hr
e.t. During the last 590 hr of operation, all three permeators lost
106
-------
and then regained flux prior to a final sodium hydrosulfide flush.
This flush virtually restored first stage permeator flux to the
original -D.015 slope values. Since the test uas terminated at
this point, it uas not possible to determine if the -O.D15 slope
could be maintained following the flush. Although the flush im-
proved permeator 3's flux, it did not return the flux to the orig-
inal slope values.
A post mortem analysis of permeator number 3 by the manufacturer
revealed that the fouling uas largely due, as expected, to calcium
sulfate precipitation.
In the case of the tuo first stage permeators uhose flux had
2
been restored by the final flush, an additional 214 kIM/m (31 psi)
pressure drop had been acquired during the study uhich uas not
removed. Since neither of these first stage permeators had ever
operated above 66 percent recovery, it must be assumed that iron
fouling caused the flux losses and pressure drop increase as it is
unlikely that calcium sulfate uould precipitate at that recovery
level on the Mocanaqua uater.
Operating parameters for Phase II hollou fiber at 75 and 85
percent recovery are given in Tables 38 and 39 respectively. Chemi-
cal data from 75 percent recovery operations are reported in Table k
Rejections during Phase II uere comparable to Phase I results,
i.e., in the range of 99 percent based on rau feed.
A complete log-log flux plot for the initial permeator, uihich
uas used both in Phase I and Phase II, is given in Figure 25. This
permeator operated for 2,670 hr (838 during Phase I and 1,832 during
Phase II).
Significant variations uere noted in the log-log flux curve of
the hollou-fiber unit. The slope appeared to increase as brine
flou decreased and a possible relationship may exist in the form of
a b/p flou limit although insufficient data uere available to sub-
stantiate this relationship. As brine flou rates decreased, recov-
ery and risk of precipitation increased and this may have caused
107
-------
Table 38. OPERATING PARAMETERS FOR HOLLOW-FIBER PHASE II STUDY AT
75 PERCENT RECOVERY AT MOCANAQUA, PENNSYLVANIA
Parameter
Value
Raui water feed flow, gpm
Total product flow, gpm
Brine water discharged, gpm
Unit ttiater recovery, percent
Feed pressure, psig
Feed temperature, °F
Permeatar No. 1 product flaw, gpm
Recovery, permeator No. 1, percent
2
Flux, permeator No. 1, gal/ft /day
Permeator No. 2 product flow, gpm
Recovery, permeator No. 2, percent
Flux, permeator No. 2, gal/ft2/day @ 400 psi & 77°F
Permeator No. 3 product flow, gpin
Recovery, permeator No. 3, percent
2
Flux, permeator No. 3, gal/ft /day
Length of run, hours
Date of run
6.18
4.?2
1.46
76.4
400
54.3
1.86
59.9
2.78
1.83
59.5
2.74
1.03
41.4
400 psi net & 77°F 1.92
1667
June 10 - August 28, 1971
400 psi net & 77°F
All values are means from 83 data sets.
Note: To convert flux (gal/ft2/day @ 400 psi and 77°F) to liters/
m2/day @ 2758 kN/m2 and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
108
-------
Table 39. OPERATING PARAMETERS FDR HOLLOW-FIBER PHASE II STUDY AT
85 PERCENT RECOVERY AT MOCANAQUA, PENNSYLVANIA
Parameter Value
Raw water feed flow, gpm 5.73
Total product flow, gpm 4.86
Brine water discharged, gpm 0.87
Unit water recovery, percent 8k.8
Feed pressure, psig 400
Feed temperature, °F 54.0
Permeator No. 1 product flow, gpm 1.89
Recovery, permeator No. 1, percent 65.5
Flux, permeator No. 1, gal/ft/day @ ^00 psi net & 77°F 2.84
Permeator No. 2 product flow, gpm 1.85
Recovery, permeator No. 2, percent 65.0
Flux, permeator (Mo. 2, gal/ft2/day § <+00 psi net & 77°F 2.78
Permeator No. 3 product flow, gpm 1.12
Recovery, permeator No. 3, percent 56.3
Flux, permeator No. 3, gal/ft2/day @ ^00 psi net & 77°F 2.10
Total operating time @ 85 percent recovery, hours 147
Longest run, hours 111
Date of run 6/26-6/28/71, 7/5-7/9/71
All values are means for 11 data sets.
Note: To convert flux (gal/ft2/day @ <+DO psi and 77°F) to liters/
rj O
m /day @ 2758 kN/m and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
109
-------
Table 40. CHEMISTRY ANALYSES FOR HOLLOU-FIBER STUDIES AT MDCANAQUA, PENNSYLVANIA
Sample
Designation
(a)
Raw feedva'
Brine(a)
Product(a)
Rejections
Raw feed
Brine
P Product
Rejections(b)
All units are
a Analyses by
Rejections
pH
3.4
2.9
4.5
3.4
3.0
4.3
mg/1
Rex
= Raw
Acid-
Cond. ity
_ —
_ _
— _
_
1020 210
3400 720
32 32
96.9% 84.8%
Cal- Magne-
cium slum
Phase I @ 75
110 80
420 310
0.55 0.59
99.5% 99.3%
Phase II 6
150 115
590 410
1.2 1.4
99.2% 98.8%
Total Ferrous
iron
Percent
65
260
0.54
99.2%
iron
Recovery
59
230
0.52
99.1%
Alum-
inum
( I C 1 C •)
8.0
31
0.18
97.8%
Sul-
fate
740
2700
2.2
99.7%
TDS
1250
4900
28.
97.8%
Manga-
nese DO
14
53
0.08
99.4%
Sil-
ica
10
43
0.49
95.1%
1 75 Percent Recovery
110
440
1.2
98.9%
except for pH and specific conductance
Chainbelt. All
other analyses
by EPA.
71
300
0.76
98.9%
15
58
0.80
94.7%
940
3000
4.6
99.5%
1320(a)
5810(a)
2.5(a)
98.1%
14(a)<1.0(
57(a)<1.0(
0.12(a)<1.0(
99.1%
a) n
a) 49
a) 0.83
(a)
(a)
(a)
92.5%
(micromhos/cm) .
feed concentration-product concentration y inn
Raw feed concentration
-------
the observed slope changes. Whatever the reason, flux slopes were
minimized when the brine flow rate uas in excess of 3.785 1/min
(1 gal/min).
Tubular Unit15'16 Phase I Studies
The tubular RO system uas also operated in tujo separate phases
during this study. Phase I utilized 60 Type 310 modules in a
6-4-2 array with five modules in each series in each rou. The last
two modules in bank 2 and the last 3 modules in bank 3 contained
volume displacement rods (WDR) that increased the brine velocity.
The normal inside diameter in the tubular system is # in (1.27 cm)
ujhich corresponds to a linear brine velocity of 1.64 ft/sec per gpm
of brine flou (0.132 m/sec per liter/min). The volume displacement
rods effectively increased this velocity to 2.5 ft/sec per gpm
(D.2 m/sec per liter/min) of brine flou. When utilizing UDR's, head-
loss through the module increased significantly. In a module uithout
l/DR's, the' headless at one gpm (3.785 1/min) brine flou uas five psi
2
(34.5 kl\l/m ) per module, while with UDR's this increased to 22 psi
(151.7 kl\l/m2) per module.
2
Initial Phase I operation uas at 6DD psi (4138 k(\)/m ) and 75 per-
cent recovery. During Phase I, the product water flux decreased
steadily from 13 gal/ft2/day to 8.5 gal/ft2/day (345 l/m2/day) in
only 48D hours. It uas originally assumed that the system had been
contaminated uith iron oxidizing bacteria because of trouble experi-
enced with the LJW light. At 16D hr, the system uas disinfected uith
a quatenary ammonium compound (L-ll-X), and n slight increase in flux
uas noted. However, membrane relaxation probably occurred during
this time since the unit operated at lou pressure. Relaxation
generally results in a flux increase for a short period of time.
Immediately after the disinfection, the flux continued to decline
rapidly. flt this point, it uas felt that the brine velocities might
possibly be too lou and that concentration polarization effects uere
causing the rapid fouling. Minimum brine velocities uere increased
from 1.2 - 1.4 ft/sec to 2.0 - 2.2 ft/sec (0.37 m/s - 0.67 m/s).
Ill
-------
Phase I chemical data are presented in Table 41.
Tubular Unit Phase II Studies
At the end of Phase I, it uas noted that the flux declines ex-
perienced with the tubular system uere not experienced with the hollou-
fiber system or the spiral-uound system also operating alongside.
It uas also noted that both the spiral-uound system and the hollou-
fiber system had considerably louier salt passage (higher rejection)
as compared to the tubular system. The high salt passage of the
tubular 31D membrane may hav/e had some influence on the flux de-
clines experienced. To test this, five high-flux—louj-salt-passage
modules (type E61D) uere installed and put into operation for Phase
II. During Phase II, recovery uas limited to less than 50 percent
o
at 4138 kl\l/m (600 psi) due to the use of only five modules. Operat-
ing parameters for this 807 hr study are given in Table 42 and
chemical data are presented in Table 41. Rejections uere higher for
Phase II (type E610) membranes than for Phase I membranes (type 310).
Figure 25 presents the flux history for these modules. An extremely
high initial compaction uas experienced during the first feu hours
of operation. The flux then stabilized on a -0.040 log-log slope
for the first 300 hr. A gradual decline then occurred through
n
about 440 hr, at uhich time the flux stabilized at 13.7 gal/ft/day
9
(557 1/m /day) for the remainder of the study. The change in the
flux decline slope experienced from near 300 hours e.t. uas also
noticed to a lesser degree on the hollou-fiber unit and uas believed
caused by a higher-than-normal iron (III) content in the AMD. The
decline uas probably entirely due to iron fouling since operation
at 40-45 percent recovery uas uell belou the CaSO fouling range.
The modules uere flushed uith a sodium hydrosulfite solution (4 ut
percent) for 1 hr. This resulted in a dramatic increase in flux
as shoun in Figure 25. Since additional operating time uas not
available, it is not knoun hou much of this flux increase uas due
to cleaning and hou much uas due to membrane relaxation. It is
felt that a substantial gain uas accomplished since membrane relaxation
112
-------
Table 41. CHEMISTRY ANALYSES FOR TUBULAR STUDIES(15>16) AT MOCANAQUA, PENNSYLUANIA
Sample
Designation pH
Raui feed
Brine
Product
Rejections
3.4
3.1
4.2
Acid-
Cond. ity
1050 250
2400 560
46 46
95.6% 81.6%
Cal-
cium
125
330
2.2
98.2%
Magne-
sium
Phase I
92
240
1.4
98.5%
Phase II 8
Raw feed(a
Brine(a)
Product(a)
Rejections
> 3.4
3.1
4.3
—
Analyses by Rex
All other
All units
So i o r"fc i nnca
analyses
-
— —
Chainbelt.
by EPA.
110
200
0.6
99.5%
83
150
0.45
99.5%
Total
iron
Ferrous
iron
Alum- Sul- Manga-
inum fate TDS nese DO
Sil-
ica
@ 75 Percent Recovery
78
230
0.9
98.8%
61
150
1.0
98.4%
12 660 1320(a) 14(a)
30 1650 3520(a) 39(a)
1.0 4.4 53(a) 0.31(a) -
91.7% 99.3% 96.0% 97.8%
11
16
7.2
34.5%
1 42.5 Percent Recovery
70
130
0.4
99.4%
are mg/1 except for pH and specific conductance
Raw feed concentration-product concentration
63
160
0.62
99.0%
8.3 800 1320 14
15 1450 2300 25
0.15 2.0 25 0.08
98.2% 99.8% 99.6% 99.4%
12
21
1.0
91.7%
(micromnos/cm ) .
x inn-
Raw feed concentration
-------
Table 42. OPERATING PARAMETERS FOR TUBULAR PHASE II STUDY AT 43
PERCENT RECOl/ERY<15,16) AT MOCANAQUA, PENNSYLVANIA
Parameter Value
Raw water feed flow, gpm 1.48
Product water flow, gpm 0.63
Brine water discharged, gpm 0.65
Water recovery, percent 42.5
Feed pressure, psig 617.0
Feed water temperature, °F 55.0
Unit flux, gal/ft2/day 8 77°F & 600 psi net 15.60
Length of run, hours 807
All values are means from 34 data sets.
Note: To convert flux (gal/ft2/day @ 600 psi and 77°F) to liters/
m2/day 9 4138 kN/ro2 and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.1,45.
114
-------
alone would not account for an increase in flux of about 33 per-
cent.
The salt passage properties for both the types 310 and 610
tubular modules are shown in Table 43. The salt passage uas cal-
culated based on average brine concentration experienced on the
membrane, i.e., the average of feed and brine concentrations.
This procedure allows comparison of salt passages while operating
the RO system at different product water recoveries.
Generally, salt passage for the type 310 modules ranged from 1
to 1.5 percent for Ca, Mn, Fe, Al, and SO,. Silica passage was
extremely high at an average percent of 54. The type 610 modules
had significantly lower salt passage; i.e., 0.4 to 0.5 percent for
Ca, Mg, Mn, Fe, Al, and SO.. Silica passage was also considerably
lower than the 310 modules at an average percent of 6.2 percent
passage. No apparent changes occurred in the salt passage through-
out the operational period.
The mechanical operation of the tubular system was excellent.
No module failures were experienced over the entire 2,800 hr opera-
tion. This reflected the improvements made in tube construction
^
since the 1969 study.
In summary, a number of points can be made regarding tubular
system operation. The use of volume displacement rods is definitely
not recommended. The price paid in headless far exceeds the bene-
fits obtained. If higher velocities are required, it appears re-
circulation of brine would be the preferred alternative. With
regard to required velocities, it appears that a minimum velocity
of about 1.5 ft/sec (0.46 m/s) is desirable, since operation at this
velocity with the higher-flux—low-salt-passage modules was satis-
factory. It should be noted, however, that the recoveries during
Phase II were quite low (45-70 percent), and that operation at
higher recoveries may require higher velocities to offset the con-
centration polarization effects due to higher brine concentrations.
Additional study at higher recoveries is necessary to answer this
question.
115
-------
Table 43. SALT PASSAGE FDR TUBULAR STUDIES(15fl6) AT MOCANAQUA,
PENNSYLVANIA
Parameter
Calcium
Magnesium
Manganese
Total iron
Aluminum
Silica
TDS
Ferrous iron
Sulfate
Salt Passage -
Type 310 Membrane Type
Phase I
1.45
1.11
1.21
1.19
1.40
53.6
2.34
1.15
1.27
Percent
610 Membrane
Phase II
0.43
0.36
0.45
0.45
0.39
5.88
0.87
0.51
0.20
Analyses by Rex Chainbelt.
Note: Salt passage is related to salt rejection. Whereas rejec-
tion measures a membrane's impermeability to passage of ions, salt
passage is a measure of the relative permeability to passage of
ions. Therefore, salt rejection plus salt passage equals 100 per-
cent if both are measured in respect to the same initial concen-
tration.
116
-------
It is not known if the high initial flux loss experienced
during Phase I with the type 31D modules uas specific for the
modules utilized or a result of the higher salt passage. In any
event, low-salt-passage modules are definitely recommended for
both flux and product water quality considerations.
Comparing the flux history of Phase I 31D modules to the flux
history from the 1969 field testing, significantly lower flux
declines were noted in the present study. For example, greater
than SQ percent of the original flux was lost in t+QO hr in the
previous study compared to about ^+5 percent of the original flux
in the present study. This would indicate that the pretreatment
system did have some effect on the tubular system operation.
A more in-depth report of the hollow-fiber and tubular studies
at Mocanaqua is available , and a complete discussion of system
comparisons made from the results of this Mocanaqua study is pre-
sented in the "Discussion" section of this report.
Calcium Sulfate Fouling at Mocanaqua
Very little or no calcium sulfate fouling occurred at 75 percent
recovery since both the spiral-wound and hollow-fiber log-log flux
plots had shallow slopes and were linear. In contrast, the severe
fouling observed at 85 percent recovery was due to calcium sulfate
precipitation. Therefore, at some point between 75 and 85 percent
recovery, the solubility of calcium sulfate was exceeded and pre-
cipitation occurred.
At 75 percent recovery, the brine was concentrated four times
in relation to the raw feed; at 85 percent, the brine concentration
was 6.67 times that of the raw feed. Typical calcium and sulfate
values were used to generate Table Ub.
117
-------
Table 44. CALCIUM SULFATE MOLAR SOLUBILITY PRODUCTS
Recovery
75%
80%
85%
Concen-
tration
factor
i*.
5.
6.67
Raw
calcium
mq/1
120
120
120
Feed
sulfate
mq/1
800
BOO
800
Brine
calcium
mq/1
480
600
800
sulfate
mq/1
1920
3000
5340
Molar
solubility
product
2k x 10"5
47 x 10~5
111 x 10"5
As the recovery was increased from 75 to 85 percent, the product
of molar concentrations of calcium and sulfate in the brine in-
creased from 24 x 10~ to 111 x 10* . Since no CaSO, precipitation
occurred at 75 percent recovery, the limiting concentration was in
excess of 24 x 10~ ; if the point of precipitation were miduay be-
tween 75 and 85 percent recovery, this would correspond to a solu-
ior
-5
bility product near 50 x 10 . It is felt that precipitation
occurred in the solubility product ranqe of 35 and 50 x 10
NORTON FERRIC IRON STUDIES COMPARING HOLLOW-FIBER A1MD SPIRAL-WOUND
UNITS
Following termination of the RO study at Mocanaqua, EPA returned
two of the three hollow-fiber modules to Norton. These were permea-
tors 1 and 2 of the Mocanaqua 2-1 array which had operated for 1,832
hr at Mocanaqua at recoveries ranging from 58 to 66 percent. Also
returned to Norton was the 4 K spiral-wound unit which had accumu-
lated 2,454 hours of 75 percent recovery operation at Mocanaqua.
At Norton, the hollow-fiber studies were conducted in two phases.
Phase I consisted of the two first stage permeators from Mocanaqua
which were operated in parallel. During Phase II, three new perme-
ators were operated in a 2-1 array similar to the Mocanaqua Phase II
study.
Hollow-Fiber Studies - Phase I
In the parallel array, recoveries of the individual permeators
could be independently varied. This allowed direct comparison of
118
-------
fouling trends as related to brine concentration. Since minimum
brine flows were not a problem with hollow-fiber systems, recycling
brine to maintain a minimum floui was not felt necessary by the man-
ufacturer.
Differential pressure (the pressure loss across a module) is an
indicator of fouling in a permeator. Although the normal differ-
ential pressure (AP) fcr a hollow-fiber permeator is 1G psi
2
(69 kl\J/m ), the A p of the two permeators at the beginning of the
Norton study was UB psi (338 kl\l/m ). Apparently the fouling ob-
served at Mocanaqua had not been completely removed before the
Norton study began.
Operation of the modules on Grassy Run water began at approxi-
2
mately 75 percent recovery and ^+00 psi (2758 kl\J/m ). These condi-
tions were maintained for the first 5QO hr (2,325 total hours) of
operation. During this period,AP increased from 50 to 90 psi
2
(3^5 to 621 kf\l/m ) in each permeator, flux declined rapidly, and
two sodium hydrosulfide flushes were needed to maintain performance
(Figures 26 and 27). Typical operating parameters for the permeators
during this period are given in Table ^5. The log-log flux decline
slope for the first 315 hr of the 500 hr period was -.035. In the
remaining 185 hr, the slope significantly increased to -0.^5 in per-
meator number 2. During the next 750 hr of operation the AP and flux
decline continued to increase even though the recovery had been low-
ered to 50 percent. The rate of decline was significantly greater
in permeator 2 than in permeator 1.
A sodium hydrosulfide—BIZ combination flush dramatically re-
duced the AP- Fouling continued, however, despite the 50 percent
recovery rate. Salt rejections decreased with each sodium hydro-
sulfide flush.
In attempts to control the fouling, several variables were in-
vestigated to determine their relationship to the problem. Bacteri-
ological effects were studied by LJU disinfection of the raw feed
119
-------
Q
UJ
C£
ioo4-
50-)i
T>
c
D
0)
C
O
O
D
TJ
* 1.04-
o
O)
3.04-
2.0
• PERMEATOR 1
© PERMEATOR 2
-H 1-
0.0
220
440
660
1540
880 1100 1320
ELAPSED TIME, hours
Figure 26
Flux and AP history for the hollow-fiber phase I study at Norton
-------
o
o
C
O
rv)
O
TJ
CM
O
O>
1
10
-o
^
o
LOG-LOG SLOPE=-0.035
I
I
I
I
I
111 I
- aPERMEATOR 1
*. ©PERMEATOR 2
I I I
567
2 3 456789 100 2 3 456789 1000 23 4
ELAPSED TIME AT NORTON, hours
NOTE: 1 gal/ft2/day = 40.7 Iiter/m2/day
400 psi = 2759 kN/m2
77 °F = 25°C
Figure 27
Flux trends during the two permeator hollow-fiber phase I study at Norton
-------
Table 45. OPERATING PARAMETERS FDR NORTON TWO PERMEATOR HOLLOW-
FIBER STUDY AT 72 PERCENT RECOVERY
Parameter
Raw water feed flow, gpm
Product water flow, gpm
Brine water discharged, gpm
Brine water recycled, gpm
Water recovery, percent
Feed pressure, psig
Feed temperature, °F
2
Tube one flux, gal/ft /day @
2
Tube two flux, gal/ft /day @>
Length of run, hours
Date of run
Value
No. 1
2.435
1.763
.672
0
72. 4
407. S
60.2
400 psi & 77°F 2.601
400 psi & 77°F 2.576
Data thru 500
September 23 - October Ik
No. 2
2.415
1.750
.665
0
72.5
hours
, 1971
All values are means from 23 data sets.
Note: To convert flux (gal/ft2/day @ 400 psi and 77°F) to liters/
2 2
m /day @ 2758 kN/m and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
D
to kN/m2, divide by 0.145.
122
-------
uater, and calcium sulfate concentration and in-module turbulence
were studied by lowering the recovery and thus increasing the
brine floiu rate. Mane of these factors mere successful in flux
stabilization, however. Acid injection to lower the feed pH from
2.8 to 2.5 uas attempted at 1,500 hr (3,332 hr total on membrane)
elapsed time to study the possibility of iron precipitation. As
illustrated in Figures 26 and 27, acid injection uas relatively
successful in flux stabilization, thus confirming the probability
of iron fouling. However, acid was so high that the supply of acid
at Norton was exhausted in 2 weeks. It required 0.0001 gal of acid
per gallon of water treated to lower the influent pH from 2.8 to
2.5 (0.12 ml/1 treated).
Since spiral-wound units had performed successfully on the
Norton water in the past, it was difficult to understand the inabil-
ity of the hollow-fiber system to cope with the ferric water- To
ensure that some unknown condition in the water would not also
affect the spiral system, the relatively dilapidated 10 K spiral
unit,which had not operated since June 1970, was restarted using
the same modules as were used in 1970. The 10 K unit operated
alongside the hollow-fiber unit from September through December
1971 for a total period of 1,094 hr. During this period, the 10 K
2
operated at 400 psi (2758 kN/m ) and approximately 65 percent re-
covery and gained flux in each of the five tubes. At the same time,
the hollow-fiber unit was fouling severely.
At the end of the 10 K test, a sodium hydrosulfide flush of the
spiral system failed to significantly improve flux. The increase
in flux observed over the 1,094 hr period was attributed to gradual
removal of fouling that had occurred during the 1970 neutrolosis
tests (the last time it uas operating).
After approximately 2,200 hr of operation at Norton (4,000 total
hr) tests on the two hollow-fiber permeators were discontinued when
a satisfactory pressure differential and flux could not be maintained
even at low recoveries. The difference in flux stability of the per-
meators between the Mocanaqua site and Norton is readily apparent in
123
-------
Figure 28 which is a log-lag plot including both sites. The single
significant factor in the hollow-fiber test was that acid injection
to control influent pH at 2.5 was absolutely necessary to maintain
any operation of hollou-fiber permeators on the Norton water
(pH 2.8).
The manufacturer requested one of the permeators to determine
cause of fouling. In return, they supplied three new permeators
to be installed in a 2-1 array as uas used in Mocanaqua.
Hollow-Fiber Studies Phase I_I
Testing on the 2-1 array began January 31, 1972, and continued
uith relatively stable fluxes as long as hLSO, uas injected to main-
tain a feed pH of 2.5. After 231.1 hr of operation, the pH probe
line fouled and stopped flou through the probe. The influent pH
rose from 2.5 to 2.8 and AP across permeators 1 and 2 increased
from 13 psi to 11D psi in approximately 8 hr. Flux, when corrected
for pressure losses, dropped roughly 10 percent in the same 8 hr
period. Permeator 3 uas not affected by the influent pH as, even
without acid adjustment, the pH into permeator 3 uas belou pH 2.5
since it received brine from permeators 1 and 2.
A log-log plot of the flux of each of the three permeators is
given in Figure 29.
A sodium hydrosulfide flush of permeators 1 and 2 uas moderately
effective on permeator 1, but a failure occurred in permeator 2.
The feed distribution tube, made of polyethylene, collapsed during
the flush and allowed the "0" ring seals to fail. Rau water went
directly from the feed into the brine line bypassing the membrane
altogether. The manufacturer repaired the permeator on-site,
flushed it, and restarted the run but a high pressure loss again
recurred across permeators 1 and 2. The manufacturer took both
permeators back to further investigate the failure and fouling and
supplied two new ones to continue the test. Operating parameters
for the study are given in Table ^6. Typical chemical analyses are
121*
-------
4
30
0)
c
'« 20
Q.
0
o
c 10
° 9
U_
O
O -j
rv
K 6
•- @> e
(VI b
en >~
o 4
U
CS o
— J
*"*-
"""^
O
™ 2
X
U
u_
1
—
NORTON
-f ftrtOTANAOHA Tr*;T
"~ O
^^
_ ^J
NOTE:
01
>
o
U
a>
gx
~V
3-D — D a_0— c
n
>|
^
-C
,^
^5
-i n ni
U
^
•-
*o*
fr?
J
•J
r *4D
w
0)
•^
"-
3ftr — c
TJ
01
_c
in
^
U.
i ea tor
t
0)
^-.
i°T
•*
•<»•
o
to
CM
O
Z
-C
V)
D
0
CM
1
1 gal/ft2/day = 40.7 Iiter/m2/day |
400 psi = 2759 kN/m2
77 °F = 25°C
1 1 1 1 1 I I I 1
S-l
O
o
_
Z
r-
V)
3
u-
^^S-D
CNO
"^
-T,-O r
3
- 01 n
-c « 0. r
^ tO < LL_ V)
1 1
|
.
| ,
i . i
1
1
-D ~ —
•o °
< « —
0)
-DO:
ai
a "S
c ai ~
'^ L^
E
3 ^
1 ^
•A*
i i i i i
4
3
2
i
1
9
8
7
6
5
4
3
2
10
56789 100
200 300 400 600 800 1000
ELAPSED TIME, hours
2000 3000 4000 5 67891
Figure 28
Flux history for the two first-stage hollow-fiber permeators
-------
cn
CM
cs*
10.- . o« --D
9
8
7
6
_ 5
0)
c 4
VJ
a 3
0
O
-o 2
c
a
u.
o
K
^
^^^
- LOG-LOG SLOPES X ^ 2 1 NEW FIRST STAGE
DURING FIRST 200 MRS. - £ E = PERMEATORS
TUBE 1 = -0.047 O
TUBE 2 = -0.050 J5
TUBE 3 = -0.028 ~°
OVERALL TUBE 3 SLOPE = -0.052 5
_ u
X
*-^= *=fc*='-t»K
'
NEW FIRST
STAGE PERMEATORS ^
-
^
_
J
JC
t/i
n
0
a
a>
"^— «^
"•
«
0
o
© Log-Log Slope = -0. 083 •-
:&%-• — — Q c
t (LOWER GRAPH) -
_c
- 1
vt
N 0
03 12
Q)
0)
^
V
^
™-
^
„
1 _
_
—
(Permeator Failure) "~
A
V ™
Log-Log Slope = -0.090
1 1 1 1 I 1 i 1 1 1
I 1
—
— --— » TUBE 1
-—--©TUBE 2
"TUBE 3
NOTE:
1 gal/ft2/day = 40.7 liter/m 2/day
400 psi = 2759 kN/m2
77°F = 25°C
1 1 1 I i 1 1 1 III!!!
9
8
7
6
5
4
3
2
1
9
8
7
6
5
4
3
1
10 2 3 4 5 6 7 8 9 100 200 300 4 56789 1000 2 3 4567891
ELAPSED TIME, hours
Figure 29
Flux trends for 6K hollow-fiber phase H study @ 69.7 percent recovery
-------
Table 46. OPERATING PARAMETERS FOR NORTON HOLLOW-FIBER 3-PERMEATOR
PHASE II STUDY AT 70 PERCENT RECOVERY
Parameter Value
Ram water feed flow, gpm 5.509
Total product flow, gpm 3.869
Product flou, permeator #1, gpm 1.303
Product flow, permeator #2, gpm 1.387
Product flow, permeator #3, gpm 1.179
Brine water flow through 1&2 (each), gpm 1.409
Brine water discharged (#3), gpm 1.64Q
Recovery - permeator #1, percent 48.0
Recovery - permeator #2, percent 49.6
Recovery - permeator #3, percent 41.8
Total unit recovery, percent 70.2
Feed pressure - unit, psig 400.0
Feed temperature - unit, °F 44.6
Permeator one flux, gal/ft2/day @ 400 psi net & 77°F 2.195
Permeator two flux, gal/ft2/day @ 400 psi net & 77°F 2.337
Permeator three flux, gal/ft2/day @ 400 psi net & 77°F 2.356
Length of run, hours 279.1
Date of run January 31-February 15, 1972
All values are means from 9 data sets.
Note: To convert flux (gal/ft2/day @ 400 psi and 77°F) to liters/
m2/day @ 2758 kN/m2 and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
127
-------
given in Table V7. The high hollou-fiber rejections are significant.
The manufacturer reported that manufacturing defects uere responsible
for the module failure following the sodium hydrosulfide flush.
The tuo new permeators uere installed at 279 hr elapsed time on
the unit. Careful modifications of the pH control system uere made
to insure that the pH probe uould not foul as before and operation
of the hollou-fiber system uas resumed.
Idithin 20 hr, a severe rain washed sediment from a neu mine road
into Grassy Run. An immediate flux loss occurred in the tuo first
stage^permeators. Correspondingly, the AP increased from 5 psi to
"* 2
62 psi (34.5 to 428 kl\l/m ). The third permeator uas unaffected.
The spiral-uound k K unit, operating alongside, suffered no flux
loss nor AP increase during the period.
A reverse flush uas partially successful in restoring flux and
2
reducing AP (from 62 doun to 35 psi) (428 to 241 k(\l/m ). After a
feu additional hours of operation, AP lowered to a normal value of
9 psi and flux again increased (though still not back to normal).
After a feu more hours of operation, houever, the trend reversed
and AP increased to near 70 psi and flux continued to drop. At
472.6 hr elapsed time (193.1 hr on permeators 1 and 2), testing uas
terminated as very little improvement in AP or flux had occurred in
permeators 1 and 2. Houever, permeator 3 uas still performing uell.
A one-hour reverse flou BIZ flush uas unsuccessful in reducing AP
or increasing flux in any of the permeators. The flux performance
of these permeators is included in Figure 29.
In all the hollou-fiber studies at (Morton, flux and AP stability
of the first stage permeators were unsatisfactory. Houever, the
second stage permeator performance uas significantly superior to the
first stage performance even though the second stage was receiving
uater with much higher pollutant concentrations.
Since the Phase I IMorton tests indicated acid injection uas
necessary to stabilize flux, it uas deduced that iron precipitation
128
-------
Table 47. CHEMISTRY ANALYSES FOR NORTON STUDY (1972)
pH Cond. Acidity Calcium Magnesium Iran Aluminum Manganese Sulfate
Hollou Fiber 6K Unit - Phase II
Acidified raw feed
Brine (from 1&2)
(Feed to #3)
Final Brine
Product #1
Product #2
Product #3
Rejections #1
Rejections #3
Spiral Wound
Raw Feed
Blended Feed
Brine
Total Product
Product #1
Product #2
Product #3
Total Rejections
2.6
2.3
2.1
3.8
3.8
3.7
2.7
2.5
2.4
3.6
3.9
3.9
4.0
1570
2950
5600
27
48
64
98.3%
97.8%
1800
3200
4400
55
60
55
48
98.3%
800
1520
2790
55
20
14
93.1%
99.1%
630
1370
2050
10
8
7
10
99.3%
96
200
330
<. 1
<.l
<.l
99.9%
99.9%
110
220
320
0.4
0.1
0.2
0.1
99.8%
29
56
100
<.l
<.l
<.l
99.7%
99.8%
30
68
96
0.1
0.1
0.1
0.1
99.9%
110
220
400
0.2
0.2
0.2
99.8%
99.9%
140
290
420
0.24
0.30
0.25
0.12
99.9%
36
70
120
<.l
<.!
<. 1
99.7%
99.9%
43
90
140
0.50
0.58
0.56
0.25
99.4%
4.0
7.7
13
0.2
0.2
0.2
95.0%
97.4%
3.3
8.1
14
<.01
•^.01
<.01
<.Q1
99.9%
1100
2300
2920
<1.0
7.0
7.7
99.9%
99.7%
810
1820
2800
3.5
3.5
2.7
2.7
99.8%
All units in mg/1 except for conductivity (micromhos/cm) and pH.
-------
was causing the observed fouling. This deduction was further rein-
forced in the first stage of Phase II tests when the pH probe fouled
and an immediate flux loss occurred. However, when the siltation
problem during the latter part of Phase II caused an immediate AP
increase, some additional insight to the nature of the fouling was
gained. Since 10 micron filtration preceded the RO unit, the fouling
particles were less than 10 microns in size and probably were col-
loidal in nature. Possibly, the fouling seen earlier during Phase I
was largely colloidal deposition reinforced by iron precipitation.
A first-stage permeator was opened and visually inspected at the
end of the Phase II tests. The outside of the fiber bundle, where
brine concentrations were the greatest, showed absolutely no fouling
as the fibers were extremely clean. Upon dissecting the bundle, the
fibers grew darker toward the central distributor tube through which
the ratj water enters. The fibers immediately surrounding the central
distributor were coated with a brown film resembling mud. Samples
taken throughout the bundle were sent to West Virginia University
for X-ray analyses. Significant quantities of allophane-like (poorly
crystalline) material composed of alumina and silica were present on
17
the membrane. The alumina/silica ratios suggested the existence
of significant quantities of clays. This result confirmed the diag-
nosis of colloidal fouling. Also present on the membranes were
large amounts of metallic iron particles which were attributed to
the carbon steel connecting rod of the Moyno high-pressure pump.
The connecting rod sheared at the end of the hollow-fiber test.
Dr. Smith of West Virginia University also investigated the use of
a combination ultrasonic and Calgon cleaning technique for removal
of the residue on the membrane. Calgon served as an effective
wetting agent and the ultrasonic treatment completely removed the
coating from the membrane surface. However, it is not known whether
this technique would be applicable in a module which had not been
dissected.
The difference in hollow-fiber flux stability at Mocanaqua and
the instability at Norton is attributed to the fact that Grassy Run
130
-------
at (Morton is a surface stream subject to siltation and sewage loads
not seen in the underground mine discharge at Mocanaqua. Even though
ferric iron was observed to be more difficult to treat than ferrous,
it is felt that acid injection to approximately pH 2.5 ujill control
iron precipitation.
The inability of the hollou-fiber system to tolerate colloidal
or particulate matter less than 10 microns in size must be deemed a
major disadvantage and would significantly limit the application of
hallow-fiber systems on acid mine streams unless the problem can be
overcome.
Spiral-Wound Studies
Although the U H spiral system was returned to Norton in Novem-
ber 1971, logistical problems delayed resumption of testing until
January 3D, 1972. This 2 mo down time allowed considerable relaxa-
tion to occur in the membrane which had operated for 245^ hr at
Mocanaqua. Consequently, when testing began at Norton, the flux
2 2
rates were in the range of 12 gal/ft /day (^90 1/m ) (higher than
at the end of the Mocanaqua study). In addition to membrane relaxa-
tion, corrections for temperature and osmotic pressure at the new
site may have resulted in slightly higher flux values.
The osmotic pressure-conductivity relationship for the Norton
water is shown in Figure 30.
Operation of the b K spiral system at Norton was at UOO psig
(2758 kl\l/m2), 70 percent recovery, and better than 10:1 b/p flow
ratio. The unit accumulated an additional 19<+6 hr of operating
time to bring the total membrane time to ^00 hr.
Flux performance is shown two ways in Figure 31: in the top plot,
the complete flux history for tubes 1 and 3 is shown; in the lower
plot, operating time at Norton is presented independently. A least
squares regression analysis provided a log-log line of best fit for
the Norton operation with a slope of -0.036 for tube 1 and -0.016 for
tube 3. Tube one's slope, though very reasonable, was significantly
131
-------
VjJ
ro
._ 30
20
in
to
LU
tt
a.
u
CO
O
10
OP = 0.308 C
.498
ACID INJECTION, OP = 0.185 C
.498
I
I
1
I
cs
200
150
100
50
Of.
ID
co
CO
LU
Of.
Q_
to
O
1000 2000 3000 4000 5000
CONDUCTIVITY (C)MICROMHOS/cm
Figu re 3 0
Osmotic pressure - conductivity relationship of Grassy Run at Norton
-------
1 2 34567891 2 34567891 2 34567891
9
8
» 6
c 5
"> 4
Q. 4
O
0 3
"^
TJ
C 2
O
u-
o
10
©/ 9
8
0 7
~Q 6
cs 5
^^ j
O
D) 3
X
3 2
^J
u_
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 L
NORTON
(/)
LU
< 00 >-
- O X tl
° Z -
< S <
MOCANAQUA STUDY ^O?
o ~r
5 "~
S^O
Q
IU
SLOPE = -0.012
A A ft — A — 2S — A — & A & A £. Jl£.A iJ. JL^T^J^D,
SLOPE = -0.009 ° ° ^ W °°° °° °° °®^^1
STUDY
(SEE
BELOW)
5
o —
_l
u_
N —
CO
Of _
O
X
^^Sp
•^
- — i
-
-
NORTON OPERATION ONLY
NOTE:
1 GAL/FT2/DAY = 40.7 LITER/m2/DAY
400 psi = 2759 kN/m2
77°F=25°C
_
SLOPE = -0.036
S A y«^ A ^ £ A A A A ft r iSiAyvA A -
vj o W" w — w — w — '*' o * ' * * * fvA rVi LJ ."jifaat
0 SLOPE = -0.016 ^" U^ '-UjtS&g
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
«•
^
OTUBE1
S ATUBE3 ~
i
Isl
CO ~"
cc
O
I
1 —
L 1 1 1 1 1 1 1
9
8
7
/
6
5
4
3
1
9
8
7
6
5
4
3
2
10
56789 100 2 3 456789 1000
ELAPSED TIME, hours
Figure 31
Total 4K spiral unit operating history
5 67891
-------
greater than the -O.D10 slope observed on the same modules at
Mocanaqua. It is felt that the increase in slope uas directly re-
lated to the fact that Grassy Run is a surface stream subject to
colloidal loads far higher than the deep mine discharge at Mocanaqua.
This contrast i.,
-------
Table 48. OPERATING PARAMETERS FOR NORTON SPIRAL-tdOUND 4K STUDY
AT 69.6 PERCENT RECOVERY
Parameter Value
Raw water feed flow, gpm 5.076
Product water flow, gpm 3.531
Brine water discharged, gpm 1.545
Brine water recycled, gpm 6.577
Minimum brine/product flow ratio, ratio/module 10.6:1
Maximum brine/product flow ratio, ratio/module 21.6:1
Water recovery, percent 69.6
Recovery of blended feed, percent 30.3
Feed pressure, psig 400.1
Feed temperature, °F 59.2
Tube one flux, gal/ft2/day @ 77°F & 400 psi net 13.68
Tube two flux, gal/ft2/day @ 77°F & 400 psi net 13.32
Tube three flux, gal/ft2/day @ 77°F & 400 psi net 13.04
Length of run, hours 1946
(4400 total hours on membrane)
Date of run January 30, 1972 - May 17, 1972
All values are means from 53 data sets.
Note: To convert flux (gal/ft2/day @ 400 psi and 77°F) to liters/
m2/day S 2758 kN/m2 and 25°C, multiply by 40.67; to convert gallons
per minute to liters/second, multiply by 0.063; and to convert psi
to kN/m2, divide by 0.145.
135
-------
spiral system under both ferrous (Mocanaqua) and ferric (Norton)
conditions. The spiral system uas significantly more tolerant to
colloidal loads than the hollow-fiber system as shouin by the satis-
factory spiral system operation on the Norton water- However, the
increase in the spiral system's rate of flux decline at Norton as
compared to Mocanaqua uias directly attributed to colloidal fouling.
Importantly, the rate of spiral flux decline uias still very accept-
able and could be controlled quite easily through periodic flushing.
No decrease in salt rejection capability was noted on the spiral
unit during the entire ^00 hr of operation.
Operation of the spiral system at high b/p flow ratios (10:1)
Q
and lower pressure (^00 psi - 2758 kN/m ') produced significant im-
provements in flux stability compared to the 1970-71 3D13 hr Norton
k h study where approximately 25 percent of the flux was lost in the
first 500 hr.
POST TREATMENT OF PRODUCT WATER
In an effort to produce a product of potable quality, product
samples from the early part of the spiral testing and from the hollow-
fiber Phase II study were neutralized to pH 7.5 and analyzed. The
three critical parameters were pH, iron, and manganese. As shown in
Table i*9, neutralization effectively increased pH and removed iron
but manganese levels remained unchanged and in excess of the D.05
mg/1 limit. Neutralization to pH 9-10 would have been required for
19
manganese removal. In that case, reacidificatian would have been
necessary to reduce pH to acceptable limits of pH 6.0 to 8.5. A
quick neutralization to pH 10 confirmed removal of manganese to
potable standards. A more complete discussion of this problem is
presented in the Discussion, Section VI, of this report under the
subsection Significances of Rejection.
136
-------
Table ^9. EFFECT OF NEUTRALIZATION ON RO PRODUCT QUALITY
Spiral-Wound
pH
Conductivity
Acidity
Alkalinity
Calcium
Magnesium
Iron
Aluminum
Sulfate
Manganese
Data 2/8/72 -
pH
Manganese
Data 3/20/72
Product
3.8
100
58
0
0.5
0.2
0.38
0.10
7.0
0.3
Norton Study
Product
i+.O
0.06
- Norton Study
Neutralized Product
7.5
60
0
10
5.6
0.2
0.07
0.08
7.7
0.3
Spiral -Wound
Neutralized Product
10.0
0.001
Hoi low-Fiber
Product Neutralized Product
3.9
65
56
0
0.2
0.1
0.07
<0.1
5.7
0.2
7.5
55
0
5
3.2
0.1
0.005
< 0.1
8.3
0.2
All units in mg/1 except for conductivity (micromhos/cm) and pH.
-------
SECTION V/I
DISCUSSION
CALCIUM SULFATE PRECIPITATION - MAXIMUM RECOVERY PREDICTION
In each of the spiral-wound studies made at the various AMD
sites, one result was significant—the limiting factor in high re-
covery reverse osmosis operation was calcium sulfate precipitation.
Gulf Environmental Systems presented a discussion of the CaSO^
problem and suggested the following method to predict precipitation:
/Pmc = 3.0 to 4.0 uhen fouling occurs, where
V Ksp
Pmc = Product of molar concentration of calcium
and sulfate in the brine.
Pmc = (Mg/1 calcium '- 40.08) x (Mg/1 SO^ 7 96.06) x 10~6
Ksp = Solubility Product of Calcium Sulfate in distilled
water where solubility & 2000 mg/1 and Ksp = 2.16
x 10~4
Unfortunately, a broad range of 3.0 to 4.0 was not adequate in
predicting RO recovery with any degree of accuracy.
Using Gulf's approach, data from each site mentioned in this re-
port were evaluated and a judgmental estimate was made of the maximum
sustainable recovery level at each site. This engineering estimate
was based solely upon experience and was thus empirical in nature and
subject to error and disagreement.
Since the brine samples from each test were supersaturated in
calcium sulfate, precipitation occurred prior to analysis. It was
felt that using raw feed values and calculating brine concentrations
based on recovery would give more reliable values for calcium and
sulfate concentrations. For example, if the RO unit were operated
at 75 percent recovery, the brine values would be concentrated four
times in relation to raw feed values. Calculations of the brine
concentration were then made based on the assumed maximum recovery.
138
-------
The major argument against Gulf's method is that Ksp is a value
for calcium sulfate solubility in distilled water. A Ksp for highly
polluted brine uith all involved ionic competitions would be a
significantly different value than that for distilled water.
Earlier work at Gulf included formulations by Marshall, Slusher.
18
and Jones for correcting Ksp of CaSO^ for ionic strength. Although
the work was at higher temperatures and sodium chloride solutions,
Gulf felt the relationships were generally valid for RO applications.
The ionic strength (I) of a solution is defined as:
, -, 2
Mi = molar concentrations of individual ions
Zi = ionic charges
Marshall, Slusher, and Jones related ionic strength (I) to Ksp
for calcium sulfate as follows:
Ksp = 1.8 x 10"3 (I) n*75
Therefore, n second approach mas used where -JPrnc/Hsp UBS calcu-
lated using Ksp corrected for ionic strength.
Table 50 presents a summary of all tests along with estimates of
maximum sustainable recoveries at those sites and corresponding
values of v/Pmc/(2.16 x 1D~ ) and >/Pmo/Ksp (corrected).
Table 51 presents a summary of chemistry analyses at all sites.
The yPmc/Ksp corrected for ionic strength was calculated from
these values.
It was felt that the results of Table 50 provided rough limits
for the calcium sulfate threshold of /Pmc/(2.16 x 10 ) = 2.0 and
7 Pmc/Ksp =1.2 (where Ksp was corrected far ionic strength).
Computer programs were then written to calculate the maximum
predicted recovery based on analyses of the raw feed water. The
program incremented recovery, calculated brine concentrations, and
solved 7Pmc/Ksp until limits of 2.0 (Ksp = 2.16 x 1D~*) or 1.2
(Kap corrected for ionic strength) were reached.
139
-------
Table 50. REVERSE OSMOSIS RECOVERY LIMITATIONS DUE TO CALCIUM SULFATE FOULING
Actual Values
Actual
Site recovery
Norton 10K - Test #1
Norton 10K - Test #2
Norton Neutrolosis-Test 4
Norton Neutrolosis-Test 5
Morgantown
Ebensburg #1
Ebensburg #4
Norton 4K - 3000 hr
Norton 4K - Neutralized brine
4K Mocanaqua #1
4K Mocanaqua #1
4K Mocanaqua #2
4K Norton (1972)
91.2%
84.8%
55.6%(b)
54.6%
50.0%
83.6%
53.2%
72.8%
50.0%
74.8%
83.9%
74.5%
75.0%
Raw
Ca
110
110
460(b
530
190
160
120
400
130
130
140
96
feed
810
850
} 4700 (b
J 3000 (b
10900
1640
1300
1100
3100
750
800
930
1060
/ Pmc
3.72
2.21
5 3.67
5 2.84
5.27
3.73
1.07
1.47
2.44
1.36
2.20
1.55
1.40
Calculated Values
/ Pmc Estimated
V Ksp maximum /• .
(corrected) recovery <
1.74
1.29
1.66
1.59
1.87
1.89
0.85
0.94
1.55
0.99
1.33
1.05
0.93
85%
85%
30%(b)
40%(b)
<1%
75%
75%
80%
20%
80%
80%
80%
85%
Mean
./Pmc '
J 2.16
2.18
2.24
2.33
2.15
2.66
2.45
2.00
1.99
1.53
1.71
1.77
1.98
2.33
2.10
/Pmc
\/Ksp
(corrected)
1.25
1.30
1.25
1.33
1.22(C)
1.45
1.26
1.14
1.16
1.14
1.16
1.22
1.28
1.24
Maximum sustainable recovery (estimated).
Blended feed values.
Excessive at any recovery.
-------
Table 51. SUMMARY OF RAW FEED CHEMISTRY ANALYSES
Site
Norton lOK-Test 1
Norton lOK-Test 2
Norton Neutrolosis
Test #4(a)
Norton Neutrolosis
Test #5(a)
Morgantown
Ebensburg #1
Ebensburg #4
4K Norton - 3000 "hr
4K Norton
4K Mocanaqua #1
4K Mocanaqua #1
4K Mocanaqua #2
4K Norton (1972)
Actual
recovery
91.2%
84.8%
55.6%
54.6%
50.0%
83.6%
53.2%
72.8%
50.0%
74.8%
83.9%
74.5%
75.0%
Cond.
1100
1200
5200
3000
7000
1500
1180
970
3200
1080
1100
1110
960
Acidity
630
540
2700
1150
5200
380
390
440
66
240
230
220
620
Ca
110
110
470
460
530
190
160
120
400
130
130
140
96
Mq
33
37
270
120
420
54
51
39
170
88
90
110
29
Total
iron
110
80
520
210
2300
135
130
130
1.3
77
80
100
115
Ferrous
iron
3
3
10
10
1300
100
96
3
0.1
64
70
73
3
Al
35
31
190
91
320
32
30
63
4.5
12
13
14
36
Sulfate
810
850
4700
3000
10900
1640
1300
1100
3100
750
800
930
1060
Blended feed values.
All units are mg/1 except for pH and conductivity (micromhos/cm).
-------
In Table 52, results of computer predictions can be compared with
the estimated maximum recovery for each site. Maximum recovery
predictions by both methods agree very well with each other and
uith the original estimate of maximum recovery.
Since agreement between both methods ujas close, use of
7PmcX2.16 x 10~4) = 2.0 is recommended since only calcium and sul-
fate determinations are needed to compute maximum recovery. Using
2.D as the limit, the following formula was derived to enable pre-
diction of maximum recovery:
R - 100 - D.D55 (Ca) x
where R = maximum recovery (percentage)
Ca - AMD feed calcium concentration (mg/1)
SO, = AMD feed sulfate concentration (mg/1)
This method of predicting maximum recovery is felt to be accu-
rate to H^ 5 percent recovery on acid mine drainage.
COMPARISON OF SPIRAL-LIOUND, HOLLOw-FIBER, AMD TUBULAR SYSTEMS
In order to compare RO systems, it was vital that all comparisons
be made on the same basis, i.e., under the same operating conditions.
For this reason, water flux, when corrected for osmotic pressure
(which corrects for water quality and recovery) and temperature,
served as an effective unit for comparing of membrane performance.
Water flux, however, did not compensate for efficient packaging of
the membranes in the RO system, and therefore was a poor basis for
comoaring Fa/stem performance.
For this report, system performance was evaluated on the basis
of product output per cubic foot of vessel volume per unit of time.
To arrive at this factor, water flux was divided by pressure vessel
volume.
Table 53 presents a summary of observed flux and productivity
data from the Mocangqua studies where spiral-wound, hollow-fiber,
and tubular systems operated on a side-by-side basis. Tubular and
Phase I spiral-wound data—although actual operation was at
-------
Table 52. COMPARISON OF PREDICTED MAXIMUM RECOVERY UJITH EMPIRICAL ESTIMATES OF MAXIMUM RECOVERY
Norton 10K - Teat 1
Norton 10K - Teat 2
Norton Neutrolosis - Teat
-------
Table 53. COMPARISON OF WATER PRODUCTION CAPABILITIES OBSERVED DURING MOCANAQUA STUDIES
Pressure Membrane
vessel Enclosed packing
volume membrane density
f ±3 araa Ft? Ft2/ft3
Avg. flux
Total
vessel
flux/da
Output per cubic foot of
vessel volume
per
per
Spiral-wound
(Phase I) 1.13
150
(19.28 @ 600) (2892 @ 600) (2559 @ 600) (1.78 @ 600)
133 12.86 1929 1707 1.19
Spiral-wound
(Phase II) 1.13
186
165
12.31
2290
2026
-F-
.c-
Hollow-fiber
(Phase
Tubular
(Phase
II)
II)
0.
0.
65
63
1500
16.9
2308
26.8
2.
(15.60
10.
48
@ 600)
40
3720
(264 @ 600)
176
5723
(418 @ 600)
280
3.
(0.29
0.
97
@ 600)
19
(a)At 77°F (50°C) and UOO psi net pressure.
33 22
Note: To convert ft to m , multiply by 0.028; to convert ft to m , multiply by 0.093; to
convert flux values (gal/ft2/day @ 77°F and kOQ psi) to l/mZ/day @ 25°C and 2758 kN/m2,
multiply by UQ.67.
-------
kl\l/m (600 psi)—ucre also normalized to 2758 kIM/m (UQQ psi)
to enable direct comparrson uith the holloui-fiber unit.
In terms of average flux, spiral-uound membranes uere clearly
superior uith 2758 kN/m2 (^00 psi) fluxes of 12.3 gal/ft2/day for
spiral versus 10. MD fnr tubular and only 2.<+S for hollou-f iber
(500 l/m2/day vs. ^23 vs. 100.8).
Because hollou-fiber permeators packed from 8-10 times as much
membrane in the same volume as the spiral system and 88 times as
much as the tubular system (see Packing Density, Table 53), the lou
specific flux of the hollou-fiber uas made up for by greater membrane
area. Although the hollou-fiber flux rate ranged from only 12 per-
cent to 20 percent tint of the spiral system, depending on the
spiral's operating pressure, the hollou-fiber product output per
cubic foot of vessel volume uas from 2.2 to 3.3 times spiral output
and from 1*4 to 21 times tubular output, again depending upon operat-
ing pressures.
The tubular system ranked louest in both flux and output per
cubic foot of vessel volume.
In Table 5^ are the relative costs for each system. Since a de-
tailed cost analysis uas beyond the scope of this report, initial
costs for purchase of one pressure vessel complete uith membrane uere
used. Dividing the initial cost by the observed output (gallons per
day of product) yielded the initial cost-per-unit-output figures
shoun in the table. Although prices and relationships observed
were valid for the Mocanaqua study, extrapolation of these figures
for large units uauld result in large errors. Other factors such
as harduare requirements uould vary uith each application and manu-
facturer and uere not included in pressure-vessel--membrane-package—
cost figures.
A third basis for comparison uas product quality. Rejections
uere calculated bv comparing the product concentration uith concen-
-------
Table 54. RELATIVE COST COMPARISONS FROM MOCANAQUA STUDY
en
System
Spiral-wound
Phase I
Spiral -mound
Phase II
Hollow-fiber
Phase II
Tubular Phase II
Cost for one
pressure vessel
and membrane
$ 850.(20)
I 850.(20)
$1000. (21)
$ 265.(22)
Observed output (gal. per
vessel per day @ 77° F
& indicated net pressure)
2892 @ 600
2290 @ 400
3720 @ 400
264 @ 600
Initial cost
per unit out-
put (gal/day)
$0.29
$0.37
$0.27
$1.00
Note: To convert gal/day to I/day, multiply by 3.785 and to convert $/gal/day to $/l/day,
divide by 3.785.
-------
trations entering the RD unit. A summation of system rejections is
presented in Table 55. Spiral-wound membranes held a slight edge
in rejecting ability.
The previously mentioned leaks during spiral-mound Phase II
studies were apparent in unit rejections (Table 55), but tube 3
rejections were of the same level as those from Phase I spiral
studies.
Under Mocanaqua conditions, the rejection advantage of spiral
membranes would be of little significance as product mater from all
three units would require additional treatment for pH, manganese,
and iron before drinking water standards could be met.
The final basis for comparison was the decrease of productivity
with time as evaluated by log-log flux decline slopes. In Table 56
are examples of log-log slopes (derived from Figures 23 and 25 and
Table 53) chosen to simulate comparable recovery conditions. In all
these examples, recoveries in the respective tubes or units were
near or below 60 percent.
Phase II spiral-wound system again had a slight advantage in
flux decline slope over the hollow-fiber systems; however, the
hollow-fiber slope was superior to Phase I spiral performance. The
tubular system's slope, calculated from the line of best fit of the
tubular log-log flux plot, was considerably steeper than either the
spiral or hollow-fiber slope.
Extrapolation of the flux curve to predict flux values at 3 and
5 years was necessarily based on the assumption that the log-log
flux decline slope remained constant. Validity of that assumption
is certainly open to debate and only sustained operation over that
period of time would prove or disprove it. If the assumption were
valid, the flux levels at 3 and 5 years of elapsed time are given
in Table 56.
-------
Table 55. COMPARISON OF MEMBRANE PERFORMANCE AT MOCANAQUA
00
System
Spiral-wound
Phase I
Spiral-wound
Phase II
Spiral-wound
Tube 3, Phase
Holloui-fiber
Phase I
Hollow-fiber
Phase II
Tubular
Phase I
Tubular
Phase II
Flux, gal/fW
day @ 77° F &
indicated pressure
19.28
@ 600 psi
12.31
@ 400 psi
11.58
II @ 400 psi
2.32
@ 400 psi
@ 400 psi
15.6
@ 600 psi
Rejections^)
Conduc- Acid-
tivity Ity
99.2 91.7
97.6 94.6
98.7
-
96.9 84.8
95.6 81.6
96.2
Cal-
cium
99.8
99.6
99.8
99.5
99.2
98.2
99.5
Magne-
sium
99.8
98.5
99.7
99.3
98.8
98.5
99.5
Total
iron
99.8
99.0
99.7
99.2
98.9
98.8
99.4
j Percent
Ferrous
iron
99.8
98.9
-
99.1
98.9
98.4
99.0
Alum-
inum
99.2
97.6
97.7
97.8
94.7
91.7
98.2
Sul- Manga-
fate nese
98.8
99.1 99.8
99.7 99.5
99.7 99.4
99.5
99.3
96.9 99.4
Sili-
con
95.1
-
90.0
Rejection = (Influent concentration-product concentration) x 100 •• Influent concentration.
22 2
Note: To convert flux gal/ft /day to 1/m /day, multiply by 40.67; to convert psi to kN/m , divide by 0.145.
-------
(a)
Table 56. PROJECTED MEMBRANE PERFORMANCEIB' - MOCANAQUA STUDIES
Flux ®fh} Log-log Predicted
100 nrs. °' flux de- flux after
Unit qal/ft^/day cline slope 3 years
Spiral-wound Phase I
Tube 1 21.5 -0.03^ 17.8
Spiral-wound Phase II
Tube 1 13.0 -0.012 12.2
Hollou-fiber Phase II
Tube 1 2.9 -0.015 2.7
Tubular Phase II 17.2 -0.063 12.1
Predicted
flux after
5 years
17.5
12.1
2.6
11.7
Ca)
Assumptions:
1. Log-log flux decline slope remains constant.
2. Salt rejection is assumed to remain constant.
Taken from log-log graph of flux versus time.
2 2
Note: To convert gal/ft /day to 1/m /day, multiply by ^0.67.
-------
SIGNIFICANCE OF REJECTIONS
Most important of the several factors that determine reverse
osmosis product quality are rau uiater quality, recovery level, and
membrane rejection caoability.
If uater of potable quality is required without posttreatment
of the product, then significant restrictions are placed upon the
mode of operation and characteristics of the RD unit.
Iron and manganese concentrations in potable water are restricted
by the U. S. Public Health Service to 0.3 and 0.05 mg/1, respectively.
Of all commonly appearing AMD constituents, iron and manganese are
the most critical for treatment to potable standards.
Since the membrane rejection ratio is constant, product quality
is directly dependent on concentrations on the brine (concentrated)
side of the membrane. Increasing recovery serves to increase brine-
side concentrations and in turn degrades product quality. Therefore,
for every rau feed concentration, it id possible to calculate the
maximum recovery that can be obtained without the product concen-
tration exceeding the U. 3. Public Health Service limit. A family
of operating curves has been developed for iron and manganese and
may be used to approximate this maximum recovery (Figures 32 and 33).
For example, if an influent water contained 50 mg/1 of iron and 5
mg/1 of manganese, the maximum recovery to meet potable standards
would be 30 percent for the iron criteria when using 99.5 percent
rejecting membranes. The same conditions allow a 66 percent recovery
before manganese limits are exceeded.
It is not necessary for the RO unit to reduce the iron and man-
ganese to this low level since the product water must receive post-
treatment in the form of neutralization to increase the pH to an
acceptable level. RO-treated water normally has a pH less than 5,
Neutralizing to pH 7 and filtering will remove residual iron, and
thus, the iron concentrations can be kept in acceptable limits. If
the iron is in the ferrous state, it will rapidly oxidize at pH 7,
150
-------
99.9% REJECTION
95.0%
(REJECTION RATE)
I I
10
2 4 6 8 10 20 40 60 80100 200
IN FLUENT CONCENTRATION, mg/l
Figure 32
Maximum operating conditions to obtain potable product
for iron limits of 0.30 mg/l
8
-------
U1
(V)
100
90
80
70
c 60
S 50
S. 40
£ 30
0% (REJECTION
RATE)
EJECTION
J
I
.2
.4
.6 .8 1.0 2 468 10
INFLUENT CONCENTRATION, mg/l
Figure 33
Maximum operating conditions to obtain potable product
for manganese limits of 0.05 mg/l
20 30 40 60 80
-------
and then it can be removed by filtering. To remove residual man-
19
ganese by neutralization, Hill states that neutralization to pH
9-10 is required. This neutralization in itself would exceed
normally acceptable pH limits and acidification of the uiater would
be needed following manganese removal.
In conclusion, assuming potable quality product water is re-
quired from the RD unit, membrane rejection is generally not sig-
nificant for iron removal since post neutralization is required
anyway. However, far AMD waters containing significant amounts of
manganese, the rejection capability is of vast importance in
achieving potable quality at minimum cost. Membranes with rejec-
tions below 95 percent would not appear desirable for AMD potable
applications.
If the RD system can be operated in a mode where the iron and
manganese can be reduced to an acceptable level by the addition of
a posttreatment step, then the unit can be operated at a high
recovery.
153
-------
SECTION VII
REFERENCES
1. Rusnak, A. and I. IMusbaum, Reverse Osmosis Field Testing on
Acid Mine Waters at Norton, tile at Uirqinia, Office of Saline
Water Contract No. 14-01-0001-1243, Gulf Environmental Systems
Report No. GA-8796, San Diego, California, August 196S.
2. Riedinger, A. B., Reverse Osmosis Field Testing on Acid Mine
Waters at Norton, West Virginia, Office of Saline Water Con-
tract No. 14-01-0001-1836, Gulf Environmental Systems Report
No. GA-91S1, San Diego, California, January 1969.
3. Gulf Environmental Systems Company, Acid Mine Waste Treatment
Using Reverse Osmosis, Environmental Protection Agency, Report
No. 14010 DYG 08/71, Washington, D. C., August 1971.
4. Rex Chainbelt, Inc., Treatment of Acid Mine Drainage by Reverse
Osmosis, Environmental Protection Agency Report No. 14010 DYK
03/70, Washington, D. C., March 1970.
5. Riedinger, A., and J. Shultz, Acid Mine Water Test at Kittanning,
Pennsylvania, Research and Development Progress Report No. 217,
Office of Saline Water, Washington, D. C., 1966.
6. E. I. DuPont de Nemours & Company, Technical Bulletin No. 100,
Wilmington, Delaware.
7. Salotto, B. V., et al., Procedure for Determination of Mine
Waste Acidity, paper given at the 154to National Meeting of the
American Chemical Society, Chicago, Illinois, 1966.
8. Environmental Protection Agency, Methods for Chemical Analysis
of Water and Wastes, 1971.
9. U. S. Steel Corporation, Sampling and Analyses of Coal and Coke
and By-products. Third Edition, 1929. -—-————-
10. Gulf General Atomic, Inc., Reverse Osmosis Principles and Appli-
cations, San Diegof California, October 1969.
11. Hill, Ronald D., R. C. Wilmoth, and R. B. Scott, Neutrolosis
Treatment of Acid Mine Drainage, paper presented at 26*6 Annual
Purdue Industrial Waste Conference, Lafayette, Indiana, May 1971.
154
-------
12. Wilmoth, R., and R. Hill, Neutralization of High Ferric Iron
Acid Mine Drainage, Federal Water Quality Administration,
Report No. li+01D ETV 08/70, Washington, D. C., August 1970.
13. Mason, Donald G., Personal communication, August 1971.
1*4. Sleigh, James H., Personal communication, August 1971.
15. Wilmoth, R. C., D. G. Mason and M. Gupta, Treatment of Ferrous
Iron Acid Mine Drainage by Reverse Osmosis, paper presented at
Fourth Symposium on Coal Mine Drainage Research, Pittsburgh,
Pa., April 1972.
16. Rex Chainbelt, Reverse Osmosis Demineralization of Acid Mine
Drainage, Environmental Protection Agency, Report No. 1^010
FQR 03/72, Washington, D. C., April 1972.
17. Grube, Walter E., and Dr. R. M. Smith, West Virginia University,
Personal Correspondence, May 17, 1972.
18. Marshall, W. L., R. Slusher, and E. V. Jones, "Aqueous Systems
at High Temperature, XIV. Solubility and Thermodynamic Rela-
tionships for CaSO^ in Nacl-H2Q solutions from <+.0° to 200°,
C, 0 to *+ Molal Nad." Journal of Chemical Engineering.
Data 9,187 (1965).
19. Hill, Ronald D., Mine Drainage Treatment, State of the Art and
Research Needs, U. S. Dept. of the Interior, Federal Water
Pollution Control Administration, Cincinnati, Ohio, December
1968.
20. Sleigh, James H., Gulf Environmental Systems Company, Personal
Communication, December 1971.
21. Potter, H. G. Jr., E. I. DuPont de Nemours, Inc., Personal
Communication, July 1971.
22. Mason, Donald G., Rex Chainbelt, Inc., Personal Communication,
December 1971.
155
-------
SECTION Will
LIST OF INVENTIONS AND PUBLICATIONS
INVENTION
A patent application (No. 213117) has been filed for the neu-
trolosis process. The title is "Reverse Osmosis - Neutralization
Process for Treating Contaminated Waters" by Ronald D. Hill,
Roger C. Uilmoth, and Robert B. Scott. As of this report, the
patent is pending.
PUBLICATIONS
The following list of presentations and publications resulted
from portions of the uork included in this report:
"Neutrolosis Treatment of Acid Mine Drainage" by Ronald D. Hill,
Roger C. Uilmoth, and Robert B. Scott, a paper presented at the
Purdue Industrial Waste Conference, Lafayette, Indiana, in May 1971;
"Treatment of Ferrous Iron Acid Mine Drainage by Reverse Osmosis"
by Roger C. Idilmoth, Donald G. Mason, and Mahendra Gupta, a paper
presented at the Fourth Symposium in April 1972;
"Mine Drainage Pollution Control by Reverse Osmosis" by Roger
C. Ulilmoth and Ronald D. Hill, a paper presented at the AIME Fall
Meeting, Birmingham, Alabama, in October 1972;
and "Mine Drainage Pollution Control Via Reverse Osmosis",
Mining Engineering. March 1973, page kS-kl, by Roger C. Wilmoth
and Ronald D. Hill.
156
-------
SECTION IX
GLOSSARY
Flux - A measure of productivity or permeability; the rate of prod-
uct flow through the membrane, usually expressed as gallons per day
per square foot of membrane area under specified conditions of tem-
perature and pressure.
Log-Log Flux Decline Slope - Even in pure water (nonfouling) systems,
reverse osmosis membranes lose flux with time. This flux loss is
linear when plotted on log-log paper. Log-log flux slopes of test
runs are indicative of the fouling that may be occurring, its sever-
ity, and its rate as compared with normal pure water values.
Pressure Drop (AP) - The pressure loss across a module or tube in
an RO unit due to hydraulic restriction in the brine channel and
piping. Increases in A P are rough indicators of fouling.
Recovery - The percentage of the raw water fed to the reverse osmo-
sis unit that results as product.
Reverse Osmosis - Flow through a semipermeable membrane where the
direction of flow is from the concentrated solution to the dilute
solution. Such a flow is induced by pressure applied to the con-
centrated solution.
Salt Rejection - A measure of a membrane's ability to selectively
allow pure water to pass through but reject the passage of impur-
ities; a measure of a membrane's impermeability with respect to
salts; usually expressed as a percentage:
(Influent Quality - Product Quality) x lnn
(Influent Quality)
Salt Passage - The percentage of salts passing through the membrane
as compared to the initial pollutant concentration. Equal to 100
minus salt rejection.
157
-------
BIBLIOGRAPHIC DATA
SHEET
1. Report^No. .
.-670/2-73-100
4. Title and Subtitle
Application of Reverse Osmosis to Acid Mine Drainage
Treatment
3. Recipient's Accession No.
5» Report Date
December, 1973
6.
7. Author(s)
Roger C. Wilmoth
8- Performing Organization Rept.
No.
9. Performing Organization Name and Address
U.S. Environmental Protection Agency
Crown MLna i)rainage Control;Field,Site, Box 555
Rivesville, W. Va. t:26588
10. Prqject/Task/Wprk Unit Nc
lBBOiio/21AFY/31
11. Contract/Grant No.
12. Sponsoring Organization Name and Address
U.S. Environmental Protection Agency
National Environmental Research Center
Cincinnati, OH lj-5268
13. Type of Report & Period
Coveted
Final Report
14.
15. Supplementary Notes
Environmental Protection Agency report number, EPA-670/2-73-100, December 1973.
16. Abstracts Spiral -wound reverse osmosis systems were tested on four different acid mine
drainage discharges in West Virginia and Pennsylvania. Comparison studies were made
of the hollow-fiber, tubular, and spiral-wound systems at a ferrous iron acid discharge;
and of hollow-fiber and spiral-wound systems at a ferric iron acid discharge.
At all sites, the limiting factor in high recovery operation was calcium
sulfate insolubility. An empirical formula was developed for predicting maximum
recovery.
Application of reverse osmosis was demonstrated to be technically feasible
for a large percentage of acid mine drainage discharges.
A process called "neutrolosis" was developed in which the reverse osmosis
brine isneutralized and clarified, and the supernatant recycled to the influent to
the reverse osmosis unit. In this manner, the neutrolosis process discharges only a
high quality product water and a neutralized sludge. Neutrolosis recoveries as high
as 98.8 percent were achieved at a ferric iron acid discharge site.
17. Key Words and Document Analysis. 17a. Descriptors
Acid Mine Drainage*
Reverse Osmosis*
Calcium Sulfate*
Coal Mines
Brine Disposal
Water Pollution Control
Iron
Manganese
17b. Identifiers/Open-Ended Terms
West Virginia*
PennsyIvani a*
Neutrolosis*
Water Recovery
17c. COSATI Field/Group
18. Availability Statement
19.. Security Class (This
Report)
UNCLASSIFIED
20. Security Class (This
Page
UNCLASSIFIED
21- No. of Pages
22. Price
}RM NTIS-35 (REV. 3-72)
-159-
USCOMM-DC 14952-P72
J.S. GOVERNMENT PRINTING OFFICE: 1974 546-516/Z63
------- |