7~-A124 244 IMPROVED FIELD PERFORMANCE FOR REVERSE OSMOSIS SYSTEMS i/I(U) NAVAL CIVIL*ENGINEERING LAB PORT HUENEME CAT A KUEPPER SEP 82 NCEL-TN-iG44

UNCLASSIFIED, F/O 13/2 1N

MEOEOEEIh'I-hEE

i '

1"1

1 RO IT";MICROCOPY BRESOUION TSTACHART96-

t4Th? A L B U E U f "NO

R O . 9"-

t]J , .

:J" t:.'-

!.,'-'

- TN NO: N-1644 0I/2-

.Lr TITLE: MPROVED FIELD PERFORMANCE FORREVERSE OSMOSIS SYSTEMS

r%

L AUTHOR: T. A.. Kuepper

DATE: September 1982

SPONSOR:AD~ Naval Facilities Engineering Command

1.1 ] UoWWU4 Marine Corps Development and Education Command

PROGRAM NO: CF60.536.091.O1.M51A

> NAVAL CIVIL ENGINEERING LABORATORY uPORT HUENEME, CALIFORNIA 93043 DTI I CoAppromed for public rem;w dbutibution unlimited. E L E CT

F J FEB 9 1.983

U A

83 012 Q&A)2

..... .....v... . . .,. ,.,,, T C

E -- - 0 N. &AI

ICCI WC

E E

[ 1 lb E w...7

198 17 6 3I''''~ ~'r'IIiriI''''''1 1

I i am isaaVVI My-l 0r~l

,..%

UnclassifiedSECURITY CLASSIFICATION OF THIS PAGE (When Doe Fn t. o)

REPORT DOCUMENTATION PAGE BEFORE COIPLET G FORM

I. REPORT NUMBER GNFLUMBER-.-:... 12 GOVT. AC ESIO ooN<., O..TN-1644 DN244043 C

4. TITLE (and SubtiIe) 5 TYPE OF REPORT A PERIOD COVERED

IMPROVED FIELD PERFORMANCE FOR Not final; Mar 1980 - Sep 1981REVERSE OSMOSIS SYSTEMS 6 PERFORMING ORG. REPORT NUMBER

7. AUTHOR(.I S CONTRACT OR GRANT NUMBER(W)

T. A. Kuepper9 PERFORMING ORGANIZATION NAME AND ADDRESS 10 PROGRAM ELEMENT. PROJECT, TASK

AREA & WORK UNIT NUMBERS

NAVAL CIVIL ENGINEERING LABORATORY 62760N;Port Hueneme, California 93043 CF60.536.091.01.M51A

I I CONTROLLING OFFICE NAME AND ADDRESS I. REPORT DATE

Naval Facilities Engineering Command September 1982I3. NUMBER OF PASES

Alexandria, Virginia 22332 87tC MONITORING AGENCY NAME G AODRESS(OI dilferent from Contfrolling Ofi.e) IS SECURITY CLASS. (of this report)

Marine Corps Development and Education Command UnclassifiedQuantico, Virginia 22134 IS. DECLASSIFICATION DOWNGRADING

SCHEDULE

16. DISTRIBUTION STATEMENT ( f his Repor)

Approved for public release; distribution unlimited.

*.' 17, DISTRIBUTION STATEMENT (of the abstract entered in Block 20. If different front Repot)

I0 SUPPLEMENTARY NOTES

19. KEY WORDS (Coni... on -sr.. ide it neces...y nd Id .ntfy by, block n-b.,)

Reverse osmosis, filter, filtration, water purification, wastewater, desalination, watertreatment, ultrasonic cavitation.

20 A S6IIRACT (Continue m reverse Aide if necessary And Identufy bF block n.bor)

. .... ... The report describes two test programs: the first one involved the physical cleaning of

reverse osmosis (RO) membranes by means of flow surging and ultrasonic cavitation. Theobjective was to clean RO membranes in situ without using chemical additives. It was shownthat ultrasonic cleaning is an effective method for removing ferric oxide, calcium carbonate/sulfate scale, and bentonite clay deposits from individual pieces of RO membranes. However,ultrasonic cavitation was not effective when applied to RO membranes in a spml ou

DD IORAN" 1473 EDITION OF I NOVAS IS OBSOLETE Unclassified

SECURITY CLASSIFICATION OF THIS PAGE (When Det Entered)

, . .

V.

UnclassifiedSECURITY CLASSIFICATION OF THIS PAGE(Whw Does &,temd)

F 20. Continued

>s Iconfiguration. Flow surging proved to be an effective method for cleaning spiral-woundRO modules in the preliminary test program conducted. A second test program involvedthe evaluation of a tubular fabric filter which has the potential of replacing conventionalmixed media filters with substantial weight and filter housing area savings. During thetest program the filter removed over half of the turbidity of the feedwater used withoutany chemical additives and could be cleaned intermittently by backwashi

Libra Card NvlCvlEgneigLbrtr

IMPROVED FIELD PERFORMANCE FOR REVERSEOSMOSIS SYSTEMS, by T. A. KuepperTN-1644 87 pp illus Septembeir 1982 Unclassified

1. Water purification 2. Ultrasonic cavitation 1. CF6O.536.091.O1.M51A

'M~Te report describes two test programs: the first one involved the physical cleaning ofrvreosmosis (R)membranes by means of flow surging and ultrasonic cavitation. The

objective was to clean RO membranes in situ without using chemical additives. It was shownthat ultrasonic cleaning is an effective method for removing ferric oxide. calcium carbonate/sulfate scale, and bentonite clay deposits from individual pieces of RO membranes. However,

Iultrasonic cavitation was not effective when applied to RO membranes in a spiral-wound con-figuration. Flow surging proved to be an effective method for cleaning spiral-wound RO modulesI

in the preliminary test program conducted. A second test program involved the evaluation of atubular fabric filter which has the potential of replacing conventional mixed media filters with

Isubstantial weight and filter housing area savings. During the test program the filter removedover half of the turbidity of the feedwater used without any chemical additives and could be

cleaned intermittently by backwashing.

UnclassifiedSECURITY CLASSIFICATION OF THIS PAGEfft-e Des Fntwred)

//

47.

CONTENTS

Page

I'l6 INTRODUCTION.................. . . ... . . .. .. .. .. .....

BACKGROUND. .. ............ .............. 2

Reverse Osmosis Process .. ...... ............ 2Military RO Hardware. ...... .............. 3Chemical Requirements .. ...... ............. 4Previous Fouling Studies. ...... ............ 4Previous Physical Cleaning Investigation .. ......... 5Feedwater Pretreatment. ...... ............. 5Previous Pretreatment Investigation. .. ........... 5

PHYSICAL CLEANING. ........................ 7

Synthetic Laundry/Shower Wastewater. .. ........... 8Actual Seawater Tests. .. ............ ..... 10Ferric Oxide Suspension Tests .. ..... ......... 11Calcium Scale Tests. .. ............. ..... 12Bentonite Clay Tests .. ............. ..... 12Ultrasonic Cleaning of Stacks of RO Module Materials .. . 13Flow Surging Tests .. ............. ...... 13

PRETREATMENT EVALUATION .. ............ ....... 14

Fabric Types .. ............. ......... 15Test Apparatus Description. ..... ............ 15Analytical Methods .. ............ ....... 16Modes of Operation .. ............ ....... 16Modes of Backwashing .. ............ ...... 17Test Results .. ............. ......... 17

SUMMARY OF TEST RESULTS .. ............. ...... 20

Physical Cleaning. .. ............ ....... 20Pretreatment Evaluation. .. ............ .... 22

CONCLUSIONS . . . . . . .. .. .. .. .. .. .. .. .. .. 22

ARECOMMENDATIONS .. ............. .......... 22

REFERENCES. .. ....... .. ............ ..... 23

APPENDIX -Graphs of Pressure Versus Time for the Accession ForSequence of Filtration/Backwashing NT ~&Experiments Shown in Table 15 . . ...... .... T

11 T AV

v

. : : , ~~. - . . .- . -. ...- . . L% ' :..- ;.::L 3 - ..- , . -? 2 . . .- o

INTRODUCTION

Reverse osmosis (RO) is now a commercially feasible, widely usedtechnology. In spite of the commercialization of RO technology however,there exists a continuing need for process improvement. As an example,water production rates are relatively low for RO membranes - on the orderof 10 to 20 gal/ft2 of membrane surface/24-hour day using a clean mem-brane and clean water source. In order to compensate for this low produc-tion rate, the most popular commercial RO modules are manufactured withrelatively high packing densities. The spiral module used within theArmy and Marine Corps 600 gph Reverse Osmosis Water Purification Unit(ROWPU) is one of the high packing-density configurations. As the pack-ing density increases within the RO module, the ability to clean theentire membrane surface decreases. It is therefore more important to beaware of membrane fouling limitations and cleaning requirements whenusing the spiral-wound configuration than for other less densely packedRO modules.

Another characteristic of the RO process is the need for high pres-sures to drive the purified water through the membrane. Typical pres-sures for operating on brackish water or freshwater is 400 to 600 psiwhile, on seawater, is about 1,000 psi. A proportionately high energycost is associated along with these high pressures. Although not asgreat as the distillation process, the energy expenditure for RO is alimiting factor for portable mobile applications due to the size andweight of associated pumps, motors, and electric generators.

In order to accommodate modest water production rates and moderatelyhigh energy expenditures, RO systems must be designed to keep the mem-brane as clean as possible for as long as possible. The conventionalway of keeping membranes clean is either of two methods: the first isto administer chemicals to periodically clean adhering contaminants fromthe membrane or to add chemicals to the feedwater to keep potential con-taminants from precipitating within the RO module; the second is topretreat the feedwater to remove those contaminants known to foul ROmembranes. For fixed installations, the above two methods are acceptableand used routinely. For the portable, mobile RO unit, chemical additivesand pretreatment hardware are not as acceptable as for fixed installa-tions and in some cases can prohibit true mobility due to chemical logis-tic trains and size/weight of pretreatment equipment.

Chemicals are often administered continuously during RO operationto keep waterborne materials from precipitating within the RO module andcausing eventual plugging of the membrane. In addition, other chemicaladditives are used periodically (perhaps once or twice per day) to cleanand restore the membrane's water production rate. For the mobile/port-able application, supplying the chemicals in field locations can be bur-

.7.. densome at the least and can be a limiting constraint if resupply is notconsistent and reliable.

*1j,

Z-., -° % % % ° • - - -. = - o . . . . , . - . . •. . . . . . •. . ... •.•° . ....• . . •••• . •. . ° •

Conventional pretreatment equipment has been optimized to operatein fixed locations and has also been developed to minimize energy andpersonnel interactions. Although land area requirements are an importantaspect of pretreatment system design, few pretreatment systems have beendesigned to operate within minimum space. Consequently, most pretreat-ment systems are bulky and heavy and have little application in portableoperation.

Sand or mixed-media filters are good examples. They are very effec-tive for RO pretreatment while requiring modest land requirements (3 to6 gpm/ft2 ) in addition to a number of ancillary facilities, includingpumps and backwash storage tanks. The filter media usually has a speci-fic gravity in excess of 1.5, which means the sand filters are quiteheavy. For example, a typical sand or mixed-media filter contains a30-inch depth of filter media or about 800 pounds of sand for a 30-in.-diam filter housing. This weight is for the sand only and does notinclude the filter housing or any of the ancillary hardware components.

Besides the bulk of pretreatment equipment and the chemical resupplynecessary, another limiting constraint is the necessity of replacingfilter media when cartridge-type filters are used within the prefiltra-tion system. Cartridge filter designs employ disposable "throw-away"media which must be replaced when fouled. The logistic burden potentialfor this type of filter can be operationally prohibitive under highlycontaminated feedwater conditions.

The effort reported here was directed toward reducing the weightand size of the multimedia filter and the need for frequent media replace-ment of the cartridge filter within the existing prefiltration system ofthe 600 gph ROWPU. This was to be accomplished by developing a light-weight modular filter capable of replacing the existing sand and cart-ridge filters in the ROWPU. In addition, this work effort was also direc-ted toward increasing the water production rate through the RO membranebetween programmed chemical cleaning cycles. A physical cleaning tech-nique was to be used without any additional logistic resupply requirement.

Improvements anticipated include increasing the ROWPU water pro-duction rate while maximizing RO membrane life by reducing membrane inter-action with feedwater contaminants. Additional improvements includedecreasing cartridge filter replacement requirements and allowing theROWPU to operate on all feedwater sources, including highly contaminatedbrackish river and lake water sources, seawater, and laundry/shower waste-water. The added ability for the ROWPU to operate on laundry/showerwastewaters will enable military forces to recycle water in the combatbase environment, thus reducing water production requirements.

BACKGROUND

Reverse Osmosis Process

The reverse osmosis process is a pressure-driven membrane separation*:i process which employs high pressure to overcome the water feedsource's

osmotic pressure caused by soluble organic and inorganic substances.4 Water is forced through the membrane which is generally impermeable to

2

'.r -• -, . r. - . . . * .: : . . - -' i ' - ,:I - - -. i

organic chemicals and inorganic ions, thus producing a purified water-stream. Four basic RO module configurations are in commercial use:tubular, hollow fiber, plate and frame, and spiral wound. Currently,spiral-wound modules are most widely marketed while plate and frame sys-tems are the least marketed. The two major classes of RO applicationare desalination and the treatment of industrial wastewaters. For desali-nation, compact spiral-wound or hollow-fiber modules are usually pre-ferred for economic reasons, and work well since feed streams in manyparts of the world are relatively clean. With severely contaminatednatural waters and wastewaters, hollow-fiber and spiral-wound modulesare often ruled out because of their inherent tendency to foul and plug.These modules usually require substantial pretreatment to ensure accept-able fouling rates under rigorous operating conditions. Although tubularmodules are the most trouble free they are also considerably more expen-sive and require more working space than the other configurations.

As previously mentioned, the spiral-wound configuration, illustratedin Figure 1, is the module type employed within the 600 gph ROWPU. Fig-ure I also shows the specific functional components of a module. Theflat sheet components are rolled around a central tube in which the puri-fied water (or permeate) is collected. The feed channel spacer is placedbetween two layers of membrane so that the rejection or barrier surfaceof the membrane faces the spacer netting. The feed flow enters therolled-up module in a direction parallel to the permeate tube. The hydro-static pressure produced by pumping action forces water through the mem-brane and into the permeate collection material which leads to the per-meate collection tube. The permeate tube is perforated to allow passageof purified water into the tube.

Military RO Hardware

The development of the RO process into military water purificationequipment has provided a solution to the problem of procuring potablewater from sources available in a combat field environment. The ROWPUis capable of treating freshwater, brackish water, and seawater. Inaddition, the capability to treat waters contaminated with nuclear, biolo-gical, and chemical warfare agents is to be provided by the RO equipmentused in conjunction with auxiliary ion exchange and carbon adsorptionunits.

The ROWPU is capable of producing about 10 gpm (600 gph) of productwater from freshwater sources and 6.6 gpm S400 gph) of product waterfrom seawater (temperature corrected to 70 F) (Ref 1). The rate of waterproduction in the ROWPU depends upon the operating pressure, normally350 to 550 psig for freshwater and 750 to 950 psig for seawater. At thesame pressure, temperature affects the rate of flow: cold water decreasesthe flow while warmer water increases flow. However, the maximum opera-ting temperature for the ROWPU unit's feedwater is about 120 0F. Sus-tained water temperatures above this figure may damage the membraneswithin the RO modules.

3

Chemical Requirements

Four chemicals are used to maintain the ROWPU rated production lev-els. These chemicals are polyelectrolyte (polymer), sodium hexametaphos-

phate (Sodium Hex), calcium hypochlorite (chlorine), and citric acid.

Polymer is continuously fed into the feedwater in order to agglomerate

small particles into larger particles that can be removed by the pretreat-ment units prior to use of the RO modules. Sodium Hex is fed into the

feedwater in order to prevent calcium scaling. Chlorine is injected in

the RO product waterline and is used as a disinfectant to inhibit bacter-

ia and other microorganism growth in the product water storage and distri-bution system. Chlorine is introduced after the RO equipment (rather

than before) because the membranes used in the ROWPU cannot toleratedisinfecting concentrations of chlorine. Citric acid is used periodical-

ly to clean the RO membranes from fouling material by feeding citric

acid into the filtered water intake to the RO modules. Used periodic-

ally, the citric acid solubilizes fouling materials such as metal hydrox-

ides and sparingly soluble salts. A rather complete list of foulants

and corresponding chemical cleaning agents is presented in Table 1.Using the aforementioned chemical additives is the conventional

method of cleaning RO membranes in situ. They work quite well and pos-sess only two disadvantages to universal application. Chemicals require

(1) a logistic train to transport and handle the materials in bulk formand (2) trained personnel to administer the chemicals in proper concentra-tions and suitable injection points.

These disadvantages generally do not adversely impact RO systems

located at fixed installations. Although reliability of chemical suppli-ers and employees to maintain chemical additive systems is a variablewhich all managers of water purification equipment must live with, thesedisadvantages take on significant importance when a RO unit is to beused in relatively remote locations throughout the world. Dependingupon the feedwater source and its inherent fouling characteristics, inter-ruption of chemical additions for whatever reason can severely restrictRO equipment water production rates.

Previous Fouling Studies

Previous studies have been performed to investigate the nature of

fouling materials on RO membranes, mechanisms of foulant buildup, andinterrelationships between fouling and operating parameters (Ref 2, 3,4, and 5). Approaches have included examination of fouled materials bychemical and physical means, as well as studies of chemical interactionsbetween membranes and chemically active foulant solutes. Although chemi-cal interactions do occur between membranes and foulants, they are usual-ly not of the type which involve rupture and formation of chemical bonds.

[... Therefore, physical cleaning methods have been found to be effective formany applications.

A reverse osmosis physical cleaning technique able to minimize thechemical cleaning additions required to maintain realistic water produc-tion rates has tremendous application for mobile or remote location desig-nated equipment.

4

Previous Physical Cleaning Investigation

Work had been conducted at NCEL prior to the current Marine Corpsproject in which a variety of physical cleaning methods that could beused to clean RO membranes were investigated. Since it was desired togive the physical cleaning techniques the best opportunity possible, atubular RO configuration was chosen as being most easily adaptable fortest and evaluation. The physical cleaning techniques chosen for inves-tigation included: air insertion (surging and continuous), flow surgingand depressurization, and ultrasonic cavitation as well as combinations

* ..- of the different methods. Figure 2 shows the hardware configurationsused for the ultrasonic equipment; Figure 3 shows test results of theevaluation of flow surging and ultrasonic cavitation, the most successfulmethods tested (Ref 6). As can be seen in Figure 3, whenever the physi-cal cleaning technique was implemented, the water production rate, orpermeate flux, increased in the two RO modules.

Testing began with a chemical cleaning to differentiate its resultsfrom that of previous tests. During the day-long evaluation, the perme-ate flux decline was modest, maintaining a high level of water productionand not requiring chemical cleaning at the end of the day. This type ofresult is exactly what was desired for the ROWPU modules although it wasrealized that the ROWPU spiral wound configuration was substantiallydifferent from the previously tested tubular configuration. This reportis the continuation of this initial reverse osmosis physical cleaninginvestigation.

Feedwater Pretreatment

Pretreating the feedwater before insertion into the RO modules isanother method for reducing the water production rate decline. This isa preventive maintenance procedure where known foulants are either re-moved or rendered innocuous before they can contact the RO membranes.In previous work, contaminants potentially harmful to RO membranes wereidentified and categorized for seawater, brackish water, and laundry/show-er wastewater (Ref 7); these data are presented in Tables 2, 3, and 4,respectively, and a summary shown in Table 5. In addition, Table 6 showsseawater composition for a variety of worldwide locations identifiedduring previous studies.

Previous Pretreatment Investigation

The initial objective of the pretreatment process investigation wasidentification of hardware or technology that could augment the existing

* pretreatment equipment of the ROWPU. The intention was to identify equip-ment that could be used upstream of the ROWPU's pretreatment system forthose locations where additional treatment of the influent waterstreamwas necessary. In addition to natural water applications, it was recog-nized that the existing ROWPU pretreatment system was not adequate tooperate in a recycling mode using 1%undry/shower wastewater as the feedsource. Therefore, it wf- desirab' that the new processing equipmentalso treat laundry/showe, tew .r, allowing wastewater to be handledby the ROWPU pretreatment "--ipment.

5

Twenty candidate pretreatment processes were evaluated on paper inrelation to the aforementioned source-water contaminants (see Table 7).From this list, three combinations of equipment or systems were chosenfor consideration, as shown in Figure 4; only system C has the potentialfor lightweight, compact packaging (Ref 8).

The roughing filters identified in systems A and B consist of coarse-media depth filters identical in principle to the multimedia filter pre-sently used in the ROWPU pretreatment system. The difference betweenthe filters is that the roughing filter media has relatively large parti-

* cle sizes to remove larger contaminants than a conventional multimedia. filter would remove. The two depth filters, therefore, if used together,

would complement each other.System C is identified as using what is termed a cross-flow filter.

Unlike the previously discussed depth filter, a cross-flow unit is asurface filter utilizing only its surface to separate water and contami-nants. Unlike the depth filter which can use its media volume to performthe required separation, a surface filter must rely solely on the squarefootage of its surface to provide adequate separation performance.Although surface filters have much greater surface area requirementsthan depth filters, the surface filter media can be (and is) lightweightand is designed in relatively high packing density configurations tomaximize surface area. The inherent operational advantage of the surface

* :filter compared to the depth filter design is that the surface filtercan be subjected intermittently or continually to rather simple physicalcleaning techniques in order to clean the filter surface. The cross-flowtechnique employs contaminated water flow parallel to (or tangential to)the surface of the filter. Purified water is driven via modest pressures(<50 psig) through the cross-flow filter in the same manner as in allsurface filters. That is, purified water travels approximately perpendic-ular to the surface of the filter. The cross flow design, however,allows the bulk of the water flow (typically 50% to 90% of the influent)to aid in keeping the surface of the filter clean by continually sweepingpast the filter surface with water in the turbulent flow regime.

Concurrent with the pretreatment process evaluation, researchersfrom NCEL were developing a surface oil coalescer to remove oil and dirtfrom a contaminated watersource. The media chosen for this applicationwere lightweight cloth fabric tubes. The objective of the design workwas development of a combination surface coalescer and filter that wouldperform as well as a depth coalescer and yet be easily cleaned. Thiswork culminated in the issuance of a patent for the unique coalescer andfilter design (Ref 9).

The cross-flow filter configuration identified in the pretreatmentprocess evaluation was readily adaptable to the newly patented surfacecoalescer/filter; therefore the two designs were incorporated into amodified configuration named the Tubular Fabric Filter, or TF , reportedherein.

6

PHYSICAL CLEANING

The spiral-wound RO configuration shown in Figure I was used duringthe physical cleaning investigation. When ultrasonic cavitation wasevaluated, a cylindrical ultrasonic transducer configuration (Figure 5)was employed. All testing was conducted by Walden Division of Abcor,Inc. at their Wilmington, Mass. facility (Ref 10).

Tests were conducted using synthetic laundry/shower wastewater,samples of actual seawater collected from a location on the Atlanticcoast near Boston (Beverly Beach, Mass.), solutions containing scaleformers (calcium carbonate and calcium sulfate), and suspensions of fer-ric oxide and bentonite clay. The calcium salt solutions were used

Ilk because they simulated a concentrated batch of seawater. Bentonite coag-ulating aid was used because it has shown that it severely fouls ROmodules when there is carryover in the effluent from clarifier pretreat-ment; ferric oxide was used because it is a principal component of theever-present foulant - rust. The combination of the latter two foulants

was particularly useful for measuring cleaning efficiency on the basisof flux recovery, area cleaned, and weight reductions. RO system per-formance in treating these process fluids was monitored by measuringflow rates (feed, permeate, and concentrate) and salt concentration.Calculations were made of permeate flux measured in gallon per squarefoot per day (gfd), percentage salt rejection, and product recovery.

Several of the tests with the fouling feeds were batch-concentrationtests (where product water was not returned to the feed tank) ratherthan total-recycle tests because, generally speaking, concentrating con-taminants in the fee results in more rapid fouling of membranes. Thisis especially true in the case of scale-forming foulants, as higher con-centrations precipitate the sparingly soluble salts, which are the prin-cipal components of scale. High pressure tests were performed to deter-mine whether cavitation can be achieved at the standard operating pres-sure (600 psig) or whether depressurization to 0 psig was required. Itshould be noted that when 0 psig pressure is shown with a correspondingwater flow through the test module, actual driving pressure is approxi-mately 20-30 psig. Tests were also performed to determine whether ornot cavitation is required for cleaning. For all tests using foulingfeed, chemical cleaning of both modules waE conducted before each test.If flux decline for the module not ultrasonically cleaned was excessive(greater than 50%), then chemical cleaning during the course of testingwas performed. The chemical cleaners recommended by the membrane manu-facturer (e.g., oxidizing agents such as citric acid) were used. At thecompletion of testing, the ultrasonically cleaned modules were cut apartand from the appearance of the membrane it was determined whether theentire length or depth of the spiral-wound modules was effectivelycleaned. From inspections of this kind, recommendations were to be madefor optimal configurations of the RO system and the ultrasonic cleaningdevice and for optimal operating parameters.

7

Synthetic Laundry/Shower Wastewater

Ultrasonic Cleaning with Flow Surges. The initial investigationinto the impact of ultrasonic cleaning on a spiral-wound, RO module inclu-ded attempts to effect cleaning at voltages from 10 to 60 volts (1.3 to46.8 watts) for periods of 1 to 5 minutes. It should be noted that powervalues have been estimated from a standard plot of power versus voltagesquared, which is derived assuming the slope (inverse of impedance, 0.013S) is constant. For the data given here the relationship between power(P) and voltage (V) is P - 0.013 V2 . Power data provided in parenthesesthroughout the report were obtained in this manner. The results of thisseries of tests are presented in Figure 6. It appeared that the ultra-sonic device significantly affected the flux upon first viewing this setof data. The peaks occurring at 6.5, 9, and 10.5 hours suggest thatoperation of the ultrasonic device led to an enhanced increase in flux.To the contrary, the activation of ultrasonics at 1, 4.5, and 8.5 hourssupports the notion that flux is enhanced without the ultrasonic device.

* .' Careful examination revealed that whenever the normal operating condi-tions were altered, the flux was affected. The "flow surges" (FS) causedby shutting off the RO system during ultrasonic operation and then turn-ing the system on again appear to have a significant impact on the flux.Therefore, little can be concluded from the ultrasonic performance data;however, because of the response of the control module, it is safe tostate that severe changes in the flowrate (flow surges) tend to enhancethe flux, most likely by sweeping away some of the foulant layer.

In Figure 7, a combination of ultrasonics and various levels offlow surging was attempted; however, again no significant difference wasdiscerned between the ultrasonic module and the control. As in all ofthe tests conducted, the experimental and control modules were operatedunder similar conditions except that the control module was not exposedto ultrasonic energy.

Throughout most of the tests following, a special effort was madeto reduce the impact that operating conditions (i.e., changes in flowrate) had upon flux values in order to document the effect of ultrasoniccavitation alone.

Ultrasonic Cleaning Without Flow Surges. In Figure 8, the resultsof attempted ultrasonic cleanings at 80 volts (83.2 watts) are repre-sented. The operating conditions employed for the RO system during aLti-vation of the ultrasonic device were the same as those employed through-out normal operation (i.e., 500 psi) without depressurization. Thisprocedure was aimed at minimizing the effect of operational variationson flux values for the data obtained in Figures 8 and 9.

At 83.2 watts (or 80 volts in Figure 8), and less significantly at130 watts (or 100 volts in Figure 9), operation of the ultrasonic deviceincreased the flux. It was noted that in each instance of ultrasoniccleaning, the permeate produced by the ultrasonic module was warmer thanthe permeate produced by the control module. This is consistent withexperimental observations of heat generation from the collapse of cavi-tational bubbles. In addition to the enhanced water production due toultrasonic cleaning, the warmer permeate would have a lower viscosity;

L8

and, since less viscous fluids tend to pass through the membrane morerapidly, the flux value would be increased. Extensive experiments aimedat a more conclusive determination of the impact of ultrasonic exposureon fresh spiral-wound RO modules were planned.

Ultrasonic cleaning was again attempted when the RO system was notdepressurized, and the results are given in Figure 10. These data con-trast the results in Figures 8 and 9 and suggest that system depressuri-

' zation is required for effective cleaning.Fresh RO modules were employed for the continuation of experiments

to discern the impact of ultrasonics on flux values over treatment per-iods ranging from 8 to 20 hours. The data represented in Figures 11 and

.Z 12 were obtained by following a single approach to ultrasonic operation.During these tests, the RO system was depressurized prior to activationof the ultrasonic apparatus. The temperature of the process fluid andits flowrate through the modules remained essentially constant duringnormal operation as well as during ultrasonic activation. This constanttemperature suggests that when the RO module was depressurized duringultrasonic cleaning, turbulence rather than heat was the primary productof cavitation.

The experiments represented in Figure 11 were performed using anapproach aimed at maximizing the flux through closely spaced, intermit-tent ultrasonic operation over longer periods of system operation thathad been used previously. It was hypothesized that a preventive approachto fouling might be more effective than a remedial approach entailing"spot" cleaning conducted only after severe flux decline. From thesedata it appears that ultrasonics was only moderately effective in dealingwith this particular foulant. However, it was clearly demonstrated that

* the foulant layer that formed on the membrane was easily removed by sim-ple flow surges.

Figure 12 is included in this report to illustrate the confusingeffect that variations in pressure have upon the flux levels. Thesedata imply that a degree of cleaning can be accomplished by merely depres-surizing the RO system without flow surges since flow rates were heldconstant throughout the duration of this test.

Threshold for Module Damage. An attempt to clean ultrasonically at125 volts (203 watts) resulted in membrane failure. The operationallife-span of the two modules is plotted in Figure 13 in terms of rejec-tion versus time. The high rejection values recorded throughout the52 hours of operation during which the data in Figures 6 through 10 wererecorded, indicate that the two modules were functionally sound. Uponexposure to ultrasonic energy at 125 volts while the RO module was con-stantly subjected to 600 psig, the ultrasonically cleaned module failedto exhibit a rejection above 90%. In fact, following this test, therejection percentage declined to zero, indicating that the process fluidwas passing through the membrane untreated and that the integrity of themembrane had been altered.

A blue dye of sufficiently large molecular size that would not nor-mally pass through the membrane was circulated through the module inorder to highlight the damaged areas. The membrane backing in damagedareas was thus stained blue, while the undamaged sections remained white.Dissection of the module revealed a 1-in.-long area on one side of the

9

permeate tube where the high-temperature PVC material had been melted bythe ultrasonic energy. Blue dye had passed through the membrane, stain-ing the backing in areas located close to the melted section. Flat cell

.* tests of membrane samples excised from the module at intervals from theexterior to interior spiral layers indicated that the damage was local-ized in one area near the permeate tube. These three bits of evidencesupport the hypothesis that the focusing of energy in the longitudinaland circular center of the transducer overcame the disseminating influ-ence of the membrane, feed channel spacer and permeate collection mater-ial. Moreover, heat loss from this focal point was prevented by theinsulating action of the spiral layers. The heat generated as a resultof the collapse of cavitational bubbles raised the temperature to themelting point of the CPVC tubing. In the area where the heat damagedthe tube, it also damaged the membrane. This theory was substantiatedwhen an attempt to melt a CPVC permeate tube having no membrane, Tricot,or Vexar covering failed, even when 150 volts (292.5 watts) was appliedto the transducer.

Chemical Cleaning. Chemical cleaning of the RO modules was per-V., formed in order to establish productivity baselines. The original fluxes

were 8.4 and 13.2 gfd for modules 2 and 3, respectively. These declinedto 3.7 and 4.2 gfd over the course of operation using the fouling feed.Cit:ic acid cleaning showed little effect as the flux values only in-creased to 4.8 and 4.6 gfd. Circulation of a detergent similar to "Biz"(an enzyme-activated detergent solution) cleaned the modules, raisingthe fluxes to 6.3 and 6.6 gfd. These were considered the maximum values

for the two modules because a certain degree of fouling is irreversibleand is dependent upon the essential characteristics and interaction ofthe membrane and foulant during the course of a typical operating period.

Actual Seawater Tests

Ultrasonic Cleaning Tests. A set of fresh RO modules were fabri-cated for treatment of actual seawater. The data presented in Figures14 and 15 were obtained again by adhering to a single approach to ultra-sonic operation. During these tests, the RO system was depressurizedprior to activation of the ultrasonic apparatus. The temperature of theprocess fluid remained constant during each operational period while theflow-rate through the modules was altered only in the instance labeled"flow surge" in Figure 15.

The experiments represented in Figures 14 and 15 were performedwith an approach aimed at minimizing fouling and flux decline throughultrasonic operation at intervals spanning a 6-hour period. The datashow no significant difference between modules no. 6 and 1 (ultrasonicand control, respectively). Figure 14 illustrates again that variationsin pressure have a positive effect upon flux levels. The data implythat a degree of cleaning can be accomplished by depressurizing the ROsystem, even without flow surges. Throughout the test periods monitoredin Figures 14 and 15, no significant differences can be discerned betweenthe ultrasonically activated module and the control. During each 6-hourperiod the flux declined roughly 1 gfd for both modules.

10

-,. .

Chemical Cleaning. The ultrasonic and control modules were chemi-cally cleaned with a 2.02 (by weight) citric acid and a 0.52 (by weight)solution of Biz detergent before and after the data in Figures 14 and 15were taken.

Threshold for Module Damage. An investigation into the potentiallydetrimental effect of ultrasonic exposure on RO modules was initiated inorder to better define a maximum voltage at which the ultrasonic devicecould be operated without destroying membrane function. Module no. 2, amodule previously used as the control for experiments with syntheticlaundry/shower wastewater, was exposed to potentials of 100, 120, and150 volts. Figure 16 shows the results in terms of flux and rejectionpercentage versus time. The conditions for this experiment were almostidentical to those under which the permeate tube of a previous modulewas melted. The variable in this case was the pressure under which themodule was exposed to ultrasonics: 0 psig rather than the 600 psig usedpreviously. The flux increased with repeated exposure to high voltages,and the rejection percentage reciprocally declined, indicating that dam-aged areas of membrane allowed salt to pass through. Even so, the per-meate tube was not harmed by the ultrasonic device at higher voltagesand atmospheric pressure. This supports the hypothesis that the energyinput to the module under high pressure resulted primarily in heat genera-tion which melted the permeate tube. At atmospheric pressure the energywas dissipated in forms related to the scattering of cavitation bubblescausing localized pressure and heat throughout the process fluid.

Ferric Oxide Suspension Tests

Module no. 2, formerly used as the control in experiments with syn-thetic laundry/shower wastewater fouling and for determination of thethreshold voltage correlated to ultrasonic module damage, was employedto treat a suspension of ferric oxide in RO permeate. The impact ofultrasonics on the removal of a clearly visible layer of ferric oxidewas measured in several ways. The distinct reddish-brown color of theferric oxide made it possible to calculate cleaning efficiency visuallyby area cleaned. In some instances, significant flux increases occurredas a result of ultrasonic cleaning.

Modules no. I and 6 were employed to treat a suspension of 100 gramsferric oxide in 20 liters of RO permeate (5 g/l). As shown in Figure 17,the modules were contaminated with this feed for 50 hours of continuousoperation at the end of which several ultrasonic applications showedminimal influence on flux. Module no. 6 was dissected and examined forvisible effects of ultrasonic exposure. It was concluded that ultrason-ics was ineffective in cleaning the RO membrane during this test. Sam-ples of membrane large enough to fit on the flat-cell test system wereexcised from module no. 6. Flux values were obtained for the highlycontaminateA membrane discs before and after ultrasonic cleaning. Fig-ure 18, a graph of some of the data in Table 8, shows the impact ofultrasonic cleaning on individual samples of membrane. It is clear thatthose exposed to ultrasonics showed enhanced fluxes. Visual inspectionsof the same membrane samples used for Table 8 indicated a close correla-tion between flux enhancement and visual removal of the reddish-brown

11

'.--.- - .... .'.' .'...- -. ..--'S.--, .........-..-..-..-.................................................................... - .'- -

-.

color of the ferric oxide foulant. Figure 19 shows the results of thevisual inspections performed on the membrane samples. Thus, Figure 19shows the percentage of area cleaned as it was measured visually, whileFigure 18 corresponds with these data by providing evidence of fluxenhancement due to ultrasonic application.

Calcium Scale Tests

Tests similar to the ferric oxide series were executed using a feedconsisting of calcium carbonate and calcium sulfate in Wilmington, Mass.tap water. Flux data for spiral modules tested for 75 hours with thissolution are presented in Figure 20. As with the other fouling feeds,it was found that the application of ultrasonics to the spiral-woundmodule caused a discernible flux recovery that was approximately equalto the flux increase caused by merely depressurizing the control module.Individual samples of membranes excised from that module were cleaned byultrasonics, visually inspected and the results are shown in Table 9 andFigure 21, where it is calculated that ultrasonics cleaned 31% of themembrane area.

Bentonite Clay Tests

The ±iith type of foulant tested, bentonite coagulating aid, isrepresentative of the general category of aluminum silicate colloid (clay)foulants. The use of clay was the most successful of the foulants testedbecause of its substantial impact on the flux as well as its visibleappearance, discernible mass, and greater adherence to the membrane sur-face.

A suspension of bentonite coagulating aid (20 grams/16 liters H0)was prepared for treatment by spiral-wound module no. 4. Over an 11-hourfouling period the flux declined 30%, a loss of productivity which threeconsecutive ultrasonic activations at 125 volts for 2 minutes failed torecover.

Fouling of flat-cell samples of membrane with a similar clay sus-pension was continued to determine the efficiency of the ultrasonicdevice in cleaning membrane fouled by clay. A 0.06-inch foulant layercomposed primarily of clay, with some residual iron oxide, caused anaverage flux decline of 46%. Cleaning efficiency determined by visualinspection is detailed in Table 10 while Table 11 data show the fluxrecoveries due to ultrasonic activation.

The cleaning efficiencies calculated from visual area measurementsand calculated from initial, precleaning (fouled), and postcleaning fluxvalues are not compatible because visible membrane damage was noted onall three ultrasonically cleaned samples. Clay deposition resulted in athicker foulant layer and displayed a greater tenacity for the membrane;it consequently required potentials of 100 to 150 volts to facilitateremoval. The order of magnitude was two to three times greater than the

-* voltage required to remove iron oxide and calcium salts. Because of thehigh voltages used, postcleaning fluxes in all three samples were higherthan initial flux values (before adding foulant) because the membranewas sufficiently damaged, allowing a high percentage of the product waterto bypass the membrane.

12

Ultrasonic Cleaning of Stacks of RO Module Materials

In an effort to explain the excellent cleaning obtained for singlemembrane disks and the poor cleaning obtained for spiral-wound membranes,stacks of membrane, feed channel spacer material (Vexar), and permeate

carrier (Tricot) were tested. Figure 22 shows the basic stack. Testswere performed using groups of several 4-cm-diam pieces of membrane,each fouled with ferric oxide and bentonite clay: one group was ultra-sonically cleaned (50 volts for 2 minutes) in the vertically mountedhousing; the other group served as a control. During the cleaning period,membranes were rotated 90 degrees every 15 seconds.

Cleaning efficiency was calculated from measurements of the area ofmembrane cleaned, flux recovery, and weight reductions. To prepare mem-

.* branes for weight measurements, they were dried in an oven (105 F) for15 minutes) and desiccator (2 hours). Flux recovery from ultrasonicallycleaned membrane samples excised from spiral modules was measured on aflat-cell system.

Results of cleaning tests are given in Table 12. While 90% to 95%of the membrane surface was cleaned for individual pieces of membrane at100 volts for 2 minutes, only 5% to 7% was cleaned for membranes in onestack. When two stacks were combined, only 1% of the membrane area wascleaned. Thus, it was concluded that the presence of the conventionalmodule materials drastically interfered with the cleaning efficiency ofultrasonics.

Tests (100 volts, 2 minutes with membranes rotated 90 degrees every15 seconds) were continued using pairs of materials to determine whethera specific spiral component was responsible for the reduction in cleaningefficiency. The results presented in Table 13 show that the percentageof membrane area cleaned for membrane alone was higher (70%) than formembrane and spacer (50%) and much higher than that for membrane andTricot (10%). Also shown in Table 13, when the spacer was altered bycutting three out of every four crossbars (modified spacer), the Membraneareas cleaned increased to 60%. Thus, it appeared that the Tricot maybe the major non-membrane culprit in the lops of cleaning efficiency.

Flow Surging Tests

Flow surge tests were conducted to assess the relative impact ofunidirectional flow surging (FS), bidirectional flow surging (RD-FS),and flow surging with the permeate flow blocked (BP-FS). For preliminarytests, a freshly fabricated conventional module (no. 11) was used at 500psi, 0.2 gpm, and 75°F to treat an 18-liter batch of ferric oxide (5 g/l)and residual bentonite clay in RO product water. The flux was allowedto decline roughly 30% prior to testing each remedial cleaning procedure.

With FS, the flowrate was increased from 0.2 to 1.4 gpm, and thepressure reduced to 100 psi (the lowest possible pressure for a flowrateof 1.4 gpm). With RD-FS the position of the entire housing was invertedprior to increasing the flowrate to 1.4 gpm. By doing so, the moduleitself was not disturbed. BP-FS was tested by occluding the permeateline (at 200-psi operating pressure) prior to increasing the flow throughthe module.

13

The results are presented in Figure 23 and Table 14. The flux re-covery due to FS in one direction averaged 28% (range of 23% to 32Z).Reversing the fluid transport direction while flow surge cleaning yieldedcomparable results: an average flux recovery of 26% (range of 10% to36%). The BP-FS caused a 28% (range of 18% to 36%) flux recovery on theaverage. The net change in flux for each of the three methods was approx-imately a 37% decline.

Another conventional module (no. 12) was fabricated to serve as acontrol in further tests. The results obtained from operation of thetwo modules in parallel and under conditions similar to those cited aboveare plotted in Figure 24. FS every 2 hours (total of 15 flow surges in31 hours) was initiated through module no. 11 while flow and pressurewere reduced to zero within the control module during those flow surges.The module exposed to a severe flow surge yielded, on the average, a1.4-gfd flux recovery--almost twice the absolute increase in flux thatthe control exhibited.

Continued operation under the same conditions (Figure 25), yet withmore frequent flow surges (every hour for a total of 15 surges in 17hours) again indicated that sudden increases in the flowrate at reducedpressure does enhance flux. The flow surge caused by reduction of theflow and pressure of the control module created a slightly greater fluxenhancement than the intended flow surge to the experimental module.The control averaged 0.68 gfd increase for every flow surge, while theflow surge module averaged 0.53 gfd per flow surge. This confirmed thatdepressurizing and repressurizing the RO module without increased flowalso produces a cleaning effect by itself.

In Figure 26 the data from blocked permeate line cleaning is pre-sented. Seven periods of permeate line occlusion (some of 5-minute andothers of 15-minute duration) of the experimental module with no changein flux or pressure exhibited an impact on the flux. The change in fluxfor the control, also with constant operating pressure and flowrate, was

on the average a 0.06 gfd decline while the module whose permeate flowwas blocked exhibited an average 0.67 gfd increase. Figure 26 shows acontrast between the intended cleaning effects of the blocked permeateline and the expected flux decline of the control module. Because theflowrate and pressure were not altered, the flux of the control modulereflected the influence of gradual fouling which is generally associated

V with uninterrupted operation. Blocking the permeate line without chang-ing any other operational parameter therefore also created a slight clean-ing effect.

PRETREATMENT EVALUATION

As mentioned in the BACKGROUND section of this report, cross-flowfiltration was identified as the most viable alternative to replace multi-

. media or sand filters. The tubular fabric filter (TF2 ) was the filterdesign chosen for further study to determine its application in the cross-flow configuration. The TF2 filtration media seems a natural for port-able mobile equipment because of its lightweight materials, relativelyhigh packing density and low pressure operation. In addition, the fab-ric's tendency to flex during flow pattern and pressure changes makes it

14.', .

h I% *~1. -. . . . . ..7. . ..-. . . . .. . ...- -- - - - - - - ----"

easier to clean the fabric tubes from waterborne contaminants than otherrigid filter designs. Because of the flexibility of the tubular designit was recognized that the TF2 can be used not only in a cross-flow con-figuration but also in an internally and externally pressurized mode.These three modes of operation are depicted in schematic form in Figure27.

As shown in cross-flow operation, the contaminated water is constant-ly being flushed past the inside of the fabric tubes while purified watertravels radially outward through the tubes. Sludge removal occurs contin-uously because turbulent flow conditions push contaminants around theflow loop.

In externally pressurized operation, the contaminated water entersthe chamber surrounding the fabric tubes while purified water travelsradially inward through the tubes. Sludge is removed intermittently asneeded. The tubes are cleaned by pressurizing the clean water outletand "blowing" the contaminants off the flexing filter media. This modeof operation has also been successfully used in combination with airflotation to simultaneously incorporate both filtration and air flotationin one process and in one piece of equipment. When air flotation isused, sludge is removed continuously with the foam out the top of thecolumn, although the tubes are cleaned intermittently as described above.

In internally pressurized operation, the contaminated water entersthe inside of the fabric tubes while purified water travels radiallyoutward through the tubes. Sludge is removed intermittently by openinga valve at the bottom of the filter media and maintaining turbulent flowconditions through the inside of the tubes.

Because of this inherent flexibility of the TF2 design, all threemodes of operation were tested. In addition, three different fabrictypes were evaluated.

Fabric Types

The simplest type of fabric (Type I) was a coarsely woven polyestermaterial. The webbing was easily visible, but holes through the webbingwere not. The Type I fabric was woven into a tubular shape without aseam. The second type of fabric (Type II) woven very finely with a bare-ly perceptible webbing and a nominal 1-micron opening, was stitched andglued to form a tube. The third type of fabric (Type III) was a thickfabric coated internally with a microporous membrane, which provided thesurface used in filtration.

The fabrics were supplied to UCLA where all of the filtration test-ing was conducted (Ref 11). These three fabrics were selected in thehope of providing varying degrees of treatment, with Type I being theleast efficient, and Type III being the most efficient.

Test Apparatus Description

The filter testing apparatus is shown in Figures 28 and 29. Theunit consisted of an 8-in.-diam plexiglass column which housed one tofour fabric tubes. The fabric tubes were 1-in.-diam cut into 4-foot-longlengths, each tube containing 1.0 ft' of surface area. The apparatus

15

.- 7 -2

was contained on a single, rolling platform. Figure 30 is a schematicdiagram of the unit.

To operate the filter, water is introduced into the "tap" waterreservoir to a convenient volume. Water is then pumped by the feed pump(a Viking gear pump with variable speed DC motor) through a calibratedrotometer and check valve. Detergent solution and standard test roaddust is also injected into the tee, using a variable speed peristalticmetering pump. The concentration of contaminants introduced into thefeedwater can be controlled by the metering pump speed or by the con-centration in the contaminant reservoir. Normally 88 grams of road dust

*: and 380 grams of military detergent were added to 40 liters of water tomake "stock" contaminant solution. This synthetic laundry wastewaterflowed through a second, larger rotometer to the top of the column. Thewastewater can be injected into the center of the fabric tubes or to thecolumn. The flow split was required in order to evaluate internally, aswell as externally, pressurized operation.

Backwash warar was introduced into the feed stream prior to thelarge rotometer. Therefore, the large rotometer indicated total flow tothe filter, during normal operation as well as during a cleaning cycle.

The fabric tubes are attached to a header where four filter posi-tions were originally provided.

The bottom header was connected to an electrically operated ballvalve. The valve was also equipped with a 3/4-inch manual bypass valve.The discharge side of the ball valve was connected to a backwash collec-tion reservoir and also to a diaphragm pump. The diaphragm pump wasused for backwashing and also for externally pressurized operation.

The filtered product water was drained from the column using anoverflow device which maintained the liquid level above the level of thefilters.

Analytical Methods

The basic analytic method used during all testing to evaluate perfor-mance was turbidity. Turbidity was originally measured using a Hachmodel 2100 turbidity meter, but later a Fischer model was used. Surfacetension measurements (du Nouy method) were also made, but liquid surfacetension was virtually unchanged during filtration as was total organiccarbon measurements. Flow rates and pressure drops were also measured

during each test.

Modes of Operation

As previously described, the filter was operated in three differentmodes. The first mode was internally pressurized (called "closed-ended")during the tests; that is, both the electrically operated ball valve andthe bypass valve were closed. The entire influent had to flow throughthe filters. The second mode of operation was "cross flow," where aportion of the influent flows past the fabric and into the backwash con-tainer' The cross-flow rate is controlled by opening the bypass valve.Additionally, very high cross-flow rates can be obtained by using thebackwash pump. In the third method, externally pressurized, the influentwas introduced to the column outside the filters and forced through thefilters by the sucking action of the diaphragm pump.

16

| '.-' . .o , , . . . - , , , , . , , ,. ,, *, o . --- o- . - . . -

Modes of Backwashins

The filter was operated with three different backwashing techniques.The first technique was to cross-wash the filter with very high crossflows in the hope of scouring filtered material from the internal surfaceof the tubes. To achieve high cross-flow rates the ball valve was com-pletely opened. The cross-flow backwash was used with the internallypressurized and cross-flow filtration modes.

The second type of backwash (also used with the described operation-al modes) used the diaphragm pump to suck effluent in the plexiglasscolumn back through the fabric so that filtered material would be re-moved from the internal surface of the tubes.

The third mode was used only with externally pressurized operation.The tubes are backwashed by simply pressurizing the internal part of thetube.

Other types of backwashing or cleaning procedures were used on atrial basis. For example, flow surging and manual cleaning were period-ically attempted.

Test Results

The three modes of filtering and the three modes of backwashingwere examined in a systematic way through a series of experiments chang-ing one operating parameter at a time. When it became apparent that anoperating mode was of little value, that mode of operation was abandoned.Many of the experiments had to be performed first on a trial basis inorder to adjust flowrates and determine appropriate operating pressures.Table 15 shows the entire sequence of experiments performed. The testnumbers shown in Table 15 correspond with the test numbers on the graphsin the Appendix. The graphs in the appendix contain pressure versustime for all the tests shown in Table 15.

Cross-Flow Operation. Experiments 1, 4, 5, 6, and 8 were cross-flowexperiments using different flow rates and cross-flow backwash rates.The cross-flow method removed approximately 70% of the inlet turbidity;no trend of removal efficiency with respect to flow rate or cross-flowrate could be determined. Experiment 1 shows relatively high effluentturbidity, but this is due to the high inlet turbidity.

The filter typically produced large quantities of product water inthe beginning of operation but the production rate gradually declined asthe fabric began to plug. Eventually, both the influent and cross flowwere being directed to the backwash storage vessel. The pressure gener-ally increased to about 5 psig, but this pressure was controlled by theposition of the bypass and the backwash pump control valves. For success-ful cross-flow operation an automatic flow controller will be needed tomake the system pressure rise and force more of the influent through thefilters.

Internally Pressurized Operation. This operation allowed the systempressure to rise to any desired value, since the entire influent wasforced through the fabric. Operating pressures up to 20 pasig were testedintermittently; however, pressures above 10 psi5 were eventually avoided

17

%

to prevent deterioration of the fabric. Type I filters developed smallpin holes at high pressures, thus allowing the influent to shoot out insmall jet streams. The seams of Types II and III failed at high pres-sures. Therefore, most of the experiments were restricted to pressuresof 10 psig or less.

Experiments 2, 3, 7, and 9 through 13 were conducted in the inter-nally pressurized mode. In general, this mode of operation producedmore filtrate than cross-flow operation, but plugging still occurred.Cross-washing was used to clean the fabric but was only marginally effec-tive. The inside of the Type I fabric is shown in Figure 31.

Externally Pressurized Operation. In this mode of operation, thediaphragm pump was used to suck water through the membranes. A total ofthree experiments were made; two were trial experiments and the third isshown as test No. 14 in Table 15. Cleaning in externally pressurizedoperation proved to be very successful. The tubes used in experiment 14were old and were used in the two trial reverse-flow experiments. Thefigures in the Appendix show that the tubes were successfully backwashedtwice, restoring high filtration rate. In actuality, the tubes used forexperiment 14 had been backwashed three times in preceding experiments.

The reverse-flow operation has the following advantages: (1) thefeed pump can also be used for backwash, and (2) the backwashing rateneeds to be only as high as the feed flowrate.

The success of the externally pressurized operation may be due tothe expansion properties of the fabric tubes. In internally pressurizedoperation, the tubes were pressurized and were visibly stretched, whichincreased the pores and void spaces in the fabric. Thus, some particleswere allowed to penetrate deep into the fabric. Then, during backwashingor cross washing, when the pressure was reduced and the tube shrank,particles were bound in the fabric and effective backwashing was hindered.

Externally pressurized operation is the exact opposite with respectto expanding and shrinking. Filtration occurs when the tube is pressur-ized from the outside, causing it to shrink around the inner supportsshown in Figure 32. During backwashing the tube is expanded, allowingthe pores and voids to expand, promoting cleaning.

Externally pressurized operation was slightly less efficient thanthe other operational modes in removing turbidity. However, it shouldbe noted that the diaphragm pump operates in a pulsating fashion, causingspikes in the velocity of water through the fabric. These velocityspikes could attribute to reduced efficiency. Further testing needs tobe performed in a pressure vessel without pulsating flow characteristics.

Fabric Types. The effects of fabric type on filter performance isnot yet clear. Trial experiments with the Type II membrane showed poorresults, due primarily to leakage at the seams. Experiment 11 with theType III microporous, membrane-coated fabric showed the poorest results-- a surprising result since it should have produced the best qualityeffluent. It was later learned that the end connections of the tubeswere inadequate and that leakage occurred around the worm gear hoseclamps that held the tubes to the PVC nipples. The leakage was deter-mined by accident when the experiment was repeated with a rebuilt mani-fold containing ribbed tubing adapters. The static pressure drop

18

through the Type III fabrics was approximately 20 psig at 1.0 gpM/ft2

using the new manifold. The pressures obtained in experiment 11 werenever higher than 15 psig due to the leakage problem at the tube's endconnections.

Spongeball Cleaning. A coating of dirt collected on the surface ofthe filter tubes in tests with Type III fabric but this coating could beremoved by gently wiping with a soft cloth or sponge. Therefore, it waspostulated that spongeball cleaning could be used to restore filter flux.Spongeball cleaning has been used frequently by University of Californiaat Los Angeles researchers with tubular configured RO modules. Thefilter header was rebuilt in order to allow spongeball cleaning.

The spongeball cleaning tests were inconclusive because the seamsof the membranes ruptured at high pressures (>20 psig) allowing unfil-tered water to escape into the effluent. The sponge ball cleaning tech-nique, however, appeared to clean the inside of the tubes; a dirt film,observed during the tests, was removed by the spongeball's wiping andscraping motion.

Flux Decline Tests. To determine the improvement that fabric filtra-tion provides for RO, flux decline tests were performed. This test is amodified form of the "fouling factor" test used frequently to evaluatepretreatment techniques.

The test is quite simple and uses Millipore or Gelman 0.45-micronfilters in a vacuum filtration apparatus. Samples of both influent andeffluent are collected and filtered. The samples are filtered individ-ually, and the volume filtered as a function of time is recorded. Highthroughput is an indication of low fouling tendency and has been correla-ted to RO performance.

Figure 33 shows the results of a flux decline test using effluentcollected during externally pressurized filtration with fabric Type I.The lower curve is for unfiltered influent, and it plugs the 0.45-micronfilter after approximately 450 ml. The effluent produced by the TF2 ,however, shows only moderate plugging after 1,100 ml have been filtered.The difference is dramatic.

Comparison of TF2 and Multi-Media Filter Equipment. The promisefor fabric filtration is even more apparent when one considers the tremen-dous area and weight savings offered by fabric filters. To show thepotential savings, a series of calculations is presented. These calcu-lations are based on using a fabric filter with thirty-six 1-inch filtertubes per square foot of superficial filter area or filter housing area,which corresponds to a center-to-center tube spacint of 2 inches. Atthis spacing a filtration rate of only 0.139 gpm/ftz in the fabric filter(an extremely low loading rate) is necessary to correspond to 5 gpm/ft

2

with a multi-media filter (a moderately high loading rate for this typeof filter). It should be noted that lowering the filtration loadingrates of a filter (measured in gpm/ft2) corresponds to loner filtrationruns between cleaning cycles and results in generally better overall

performance. To illustrate the area savings of the tubular fabric filterover a sand or multi-media filter, Table 16 shows a series of typicalmulti-media filter flows and the equivalent flow needed in the tubular

19

....

fabric filter to utilize the same superficial filtration area. For exam-ple, the filter housing area used in a sand filter at 6.0 gpm/ft2 (approx-imately the ROWPU multimedia filter flow rate) could be used to producethe same net flow with a tubular fabric flow at only 0.167 gpm/ft 2 offabric. This low filtration loading rate is far below the rates used

* during this test program. In reality, a much greater space savings overa conventional sand filter will be realized because the actual loadingrate of the TF2 is between 0.5 and 1.0 gpm/ft2 . To match the TF2 actualperformance, a correspondingly sized sand filter would have to operateat a loading rate over 15 gpm/ft2 . This magnitude of flow is not possi-ble for a conventional depth filter in order to maintain filtration effi-

* . i ciency. Another way of describing the difference between the TF2 andsand filter production capacities is as follows: assuming a filtrationloading rate of 1 gpm/ft2 for the TF2 (based upon actual filtration sur-face area), a filter capacity of 36 gpm can be provided occupying only 1ft2 of filter housing area.* A multi-media filter operating at a filtra-

tion loading rate of 5 gpm/ft2 (in this case, the actual filtration sur-face area and the filter housing area is the same) would occupy 7.2 ft

2

of filter housing area to provide the same filtration capacity of 36gpm. Therefore, the TF2 provides 36 gpm/ft2 of filter housing area whilethe multi-media filter provides about 5 gpm/ft2 of filter housing area.

SUMMARY OF TEST RESULTS

Physical Cleaning

Ultrasonic Cleaning. Ultrasonic cleaning of single-layer membranedisks was successful as shown by the contrast between ultrasonic andcontrol flux values, and visual inspection as shown in Figures 18, 19,and 21 from data depicted in Tables 8, 9, 10 and 11.

Using the most highly fouling feedwaters (mixtures of iron oxideand calcium scale and/or iron oxide and bentonite clay), changes in theweight and the visible presence of foulant on the membrane surface follow-ing ultrasonic cleaning were repeatedly discernible. The obstacle tosuccessful operation of an ultrasonic cleaning technique was clearly not

' the absence of ultrasonic cavitation, but the density of the spiral-woundRO configuration.

Tests with individual pieces of membrane and stacks of module mater-ials shown in Tables 12 and 13 led to the conclusion that a spiral-woundmodular configuration is not suitable for ultrasonic cleaning if it iscomprised of standard materials of construction. The key to success fora spiral design initially appeared to be the selection of improved modulematerials which are more efficient in permitting the transmission ofultrasonic energy. Further tests revealed, however, that 20 kHz ultra-sonic waves up to 250 volts for periods of 5 minutes have little

*This is equivalent to 36-gpm filtration capacity/ft2 of filter

housing area.

20

impact on the flux of fouled spiral-wound RO modules, even when fabrica-ted with materials that had relatively high transmission rates when tes-ted individually and in stacks of the spiral-wound material components.

Flow Surging. In early attempts to produce a cleaning effect withultrasonics at voltages from 10 to 60 volts (0.13 to 46.8 watts) forperiods of 1 to 5 minutes, flow surging proved to be more effective.From the results of this series of tests, which are presented in Figure6, it appeared that the ultrasonic device significantly affected theflux. However, careful examination of the flux peaks revealed that when-ever the normal operating conditions were varied, the flux was influenced.The flow surges (FS) caused by shutting off and restarting the RO systemduring ultrasonic operation appeared to have a significant positiveimpact on the flux. Therefore, it was concluded that flow surges hadthe potential to enhance the flux of RO spiral-wound modules.

A comparison of the two modules monitored in Figure 7 clearlyreveals the fluctuations resulting from the flow surges. The data dis-play shows almost identical flux increases between the ultrasonicallyactivated module and the control module. Both test modules experiencedthe same flow surge, but only the ultrasonic module utilized ultrasonicactivation. It is evident that flow surging was the dominant membrane-cleaning mechanism during this test.

In contrast to Figures 6 and 7, the data depicted in Figures 8 and9 show that when the effect of flow surging was minimized there was noflux improvement of the control module. Modest flux enhancement in Fig-ures 8 and 9 was solely due to the ultrasonic activation.

In addition to the successful evaluation of flow surging, it wasdocumented that merely depressurizing the RO module, without causing aflow surge, creates a cleaning effect (see Figures 24 and 25) by thecontrol module's increase in flux during repressurization. Likewise,blocking the permeate line without altering any other operational para-meter also created a slight but positive cleaning effect (see Figure26).

Control Equipment for Automation of Flow-Surge Cleaning. Complete

automation of flow surging and permeate line occlusion can be accom-plished through piping and valve modifications. Figure 34 representsthe process diagram for the existing ROWPU design while Figure 35 showsa schematic of proposed modifications to the ROWPU to be implementedduring optimization testing. It is anticipated that this would be themaximum number of modifications necessary, and it is probable that afinal design would entail fewer modifications than warranted for testpurposes. Valves V., V , and V and the high pressure pump are in theexisting ROWPU piping configuralion and would not be added. Valves indi-cated as V, V , and V6 would be added if an automated testing system isimplementea. Ruxiliary low pressure pumps PI and P would be added fortesting purposes; however, in a final design all effort would be made toutilize existing pumps already in the current ROWPU configuration.

21

-7 S. -77•7..

Pretreatment Evaluation

The TF2 investigation has shown that tubular fabric filtration holdspromise for development using the cross-flow and externally pressurizedconfigurations. Although the tests were performed on an experimentalmodel test apparatus, the series of tests demonstrated that the fabriccould be successfully kept clean while operating on a severely contamina-ted synthetic waste water feedsource. The fabric filters removed wellover one-half of the turbidity without chemical additives, and betterefficiency is anticipated with more uniform filtration rates.

The potential weight and area savings for the TF2 over conventionalmultimedia filters are substantial. Due to the TF2 filter media packingdensity and vertical orientation, a unit with 36-gpm filtration capacitycan occupy the same size filter housing as a 5-gpm capacity, mixed mediafilter.

CONCLUSIONS

1. Ultrasonic cleaning is an effective method for removing ferric oxide,calcium carbonate/sulfate scale, and bentonite clay deposits from individ-ual pieces of RO membranes. Ultrasonic cleaning does not appear to beeffective, however, when applied to RO membrane in a spiral-wound config-uration.

2. Flow surging with and without permeate line occlusion is an effectivemethod for cleaning spiral-wound RO modules fouled by ferric oxide, cal-cium carbonate/sulfate scale, and bentonite clay deposits. Flow surgingis an attractive auxiliary cleaning method for many commonly occurringfoulants that are currently removed by administering chemical dispersionor solubility additives.

3. The TF2 was shown to be an effective surface filter able to becleaned intermittently when fouled. The TF2 potentially can producewater quality comparable to that produced by a mixed media filter withsubstantial savings in weight and filter housing area.

RECOMMENDATIONS

1. It is recommended that flow surge optimization testing be conductedon a full-scale spiral-wound module. When successfully completed, aROWPU should be modified to evaluate the flow surging technique on afull-scale RO system.

2. It is recommended that the tubular fabric filter be optimized forcross-flow and externally pressurized operation. Once optimized, a 20-to 40-gpm unit should be built and tested in conjunction with a ROWPUwith and without the existing multi-media filter within the ROWPU pre-treatment system.

22

REFERENCES

1. U.S. Army Mobility Equipment Research and Development Command. Tech-nical Manual: Operator and organizational support maintenance manual -600 gpa reverse osmosis water purification unit. Fort Belvoir, Va., Jul1978 (Draft)

2. G. Belfort. "Pretreatment and cleaning of hyperfiltration (reverseosmosis) membranes in municipal wastewater renovation," Desalination,vol 21, 1977, pp 285-300.

3. M. T. Brunelle. "Colloidal fouling of reverse osmosis membranes,"Desalination, vol 32, 1980, pp 127-135.

4. J. M. Jackson and D. Landolt. "About the mechanism of formation ofiron hydroxide fouling layers on reverse osmosis membranes," Desalination,vol 12, 1973, pp 361-378.

5. S. Sourirajan, editor. Reverse osmosis and synthetic membranes,theory - technology - engineering, National Research Council, Canada,1977.

6. Civil Engineering Laboratory. Contract Reort CR 78.010: Test pro-gram for physical cleaning and fouling prevention in reverse osmosissystems. Port Hueneme, Calif., Membrane Systems, W.E.T., Inc., Feb1978.

7. . Contract Report: Evaluation of pretreatment for field-oriented reverse osmosis units. Orange, Calif., PRC Toups, Aug 1980.(Contract no. N68305-79-C-0023)

8. . Technical Memorandum M-63-80-28: Field performanceimprovement for reverse osmosis units, by T. A. Kuepper and R. S. Chapler.Port Hueneme, Calif., Dec 1980.

9. U.S. Patent 4,282,097: Dynamic oil surface coalescer, by T. A.Kuepper and R. S. Chapler. Aug 4, 1981.

10. Naval Civil Engineering Laboratory. Contract Report: Ultrasoniccleaning of reverse osmosis membranes in situ, final report. Wilmington,Mass., Abcor, Inc., Walden Division, Sep 1981. (Contract No. N68305-79-C-0027)

11. . UCLA-Eng-81-41: Development of a tubular fabric fil-ter concept - Phase I. Los Angeles, Calif., University of California atLos Angeles, Sep 1981.

12. Personal ltr from Mr. Harry Goto, DRDME-GS, MERADCOM, Ft. Belvoir,Va., 31 Mar 82.

23

r.~~~ -n 4j 1- 44---~-*----~----

8 0.rv 0A 4.

0d 0. (a 00r0

0 : 1 0 cc -H:3 Wu- 0 4 -4 41 0 L.)-140

001 -H.I. 0 w 4 O 4'.U 4 >W 0. 41w d1WU4 .-c 4.1 :

0). 411 1.490 0- 44 V4

00 0 0. c*U MA02 coW .O$ 4 U "e 00)'C

0 t Co 41 G) 4 44 -4:10 .-4 Z.0)e 0. 0 44

44 0)O 845.0

o u

0 0- 0 .aj1o 02.41 u~ 'w4 0

040

0 0 1~

0 4) .I 0 0ll0cc 0 U W . 0 .0.3

C4 r-4~ 0 .

CC- 0- H 0- r

0 z 0 0 :30 - 3

"q -4 -4 0 0~'- '0 . '.4 .4) 4C4 .4 0 0TH )C 0. -H I

(A (A 0N J U0c 420 0 0' 0 0 0 v 0 vN

0 10 41 0 .40 4* V 0 1 'A4 4.1 U0 Aj -UV3- 04 4) 0 0 -H ~0 -r 0 -40 ri 0'.

C.. 0 d I 4 to4f4 .114 m .-4 m 04 0 L * o 0.0q 0 -4t- ~, 0 ' .? . N O

*d U n 0 a- 0 0 01 u .-4 01 1 0-f4 * 0 0041 4.5 to ~ -H UU "4 m "M

cc$4" 0 :~ w ,-0 w 'M 4 0 V.4 0410 4j 0H $I 4t U e~ Ai C14 4j 'o IV4 1W 4 -. W0 00.A c v H.eQ. d00.0

0.'. u0 E . -4 4 0 )z.

1W( W wi '4 0 O cU0J0 UO

- 00

0 0 0 Q 0 0Att-4 V . .,.4 *a 4 V.4 -H 1W 0 d 0

0 00 20 20 022r 4 be e e~0 l~e00 be 01 U4 0. '4ya 0 W~. 0 a $4 H 00 W 0 1Wr.

o . 0 .0 40 *0 0 0 .0 42 0 w0cc04 w w001.p1400e. P-1e 0%- 4- %O 1 4 0-4

1. a4 '- 4. -440 I-4 u 44 f. P4 C

24.

CD 0. 1.4 11441 U 41 4

0 414W~4 HV41- .0 .0

SC.)~ 0 -- 4 .00 .-4 41V

'0 4 >4 -50 0A j0c -4 r-4 1

0)H 414 -4 0 040041 4)1a) 4.4 4.

en1 Z41 W

"4 00w141 a-441 c CA > 41 40

41 ~-4. -41 1

o J0 0 mW 0 U4Q. 41 a~d 60. 0 w

w1 4. 410- r4 -4-U d U 4 U

0 0 0 *4 0 0 0U0 u Is :. w Cf04 u 0. 0

r. Enl 0 1 ). 0 0-~

0

.0 a) -$ 0 0 Aco 41 w CO *4 Ai 0 A0 ca

E-4 04 0 0 0 U d 14 01 z% z4

u- 4 0 -.. 4 '-4 -4 4100o 0 1 0 -H PO0. -. 4 0 44 0 0 .0 -.0 0 10 -H I- 4 *t C O R p-4 WO t -H 4 41 L if44

04 z W U- ~ 0 -44 0) -% 4. UH . -A03 LI u 0: -1t' 00 C *.U aA " Ln O% :3 -41 1.4 * d 41 $* 1 -4 * r0. 0 CO'.0 1 I.-4 -4 '

41 41 ) -* 0 0 0 4 4j > *-4 c.- c oO o00 0co 00-4 .0r. 0 - a 40 uI 4 W-e 0A C10 LU 4J0OOH HO -A- *0 -4 .0 00 '0 0 0 0 0 0

OHU 41 U Ok f W UUVAi 0 0. OD 01 41 VU4 Ai U VHU44

41 U 1 U n IA-' H4 U H -4 V4.0 -A H HC4 : 23 H = I u -4~ 4 IU e 4 ~ u u m: 3= iu 0 -40 0 0. a E-4 $ 0 1 0 3 AU 4 - 1 E-4 0 . .4 CL -I 1 d 0. -4-A 0 z 0 wCO 0 0u W 4 .4U 4 H Q U1 -A -%0 m0~ 10 U "uu"wUA

41 H c4100 w A r~4 k O 0 H0 0 - W M1 4H H4 M W.a 4V0 00 w.~ =. 0~ 1 H C14Q.0 0W 44 0 4 : W04

41 -H~ -n ., H " 4-4 U1 .4. 41 *0 - H1.$ ''I 0 n E-4.U -i J 1 E-4*-nAU' UO 1 r) 0 0 H '0-40 144 4O a0~ -0 p

1.4

41 to.0 cc toH

U 0 41 go 4 U0 3 41 W u 4V-9 3 9

to 0I. 411. 4.Ir 0 co 0o go '~-4 0

924 (00 n en M t -4 -1- -

go, clog ~ 0 0

25

0 C:45

0WC) *'4.4 %" ).4Z 4. ) 44

V 0 v4 Q. -4

co w) (q) M- 0 0w 0-0 U0 V . 4

r. 0 td 6 4J0 -.C:0 6 '4 41 000 1-

0 % 0 9: 04AJS 4'

'-4 A 4~ 04

14 4441r - 0 1-4C

t'. w C 4 ml 450 -I-iC.4mp4 a il.444 0 Go 40

Ad44 1 Ai4 -4 44 -4 to 44

J 4)

44 4

0 0. 00-'4 0 u.4 a

00

ON 316 0 4Aw &to 45 C41 U ."

'-04 u 0 X40.ac 0 0 0 -A On Q 0 0i

0 .0 040 45 0.4500 '0 4.5 0 ~ &n. in $6 0 044 A.4-4 ., 0- cc a-i4J I5 - .. 4.

-. 4 04n5 V 0 0 0*. 0 - 1-4 to0"0 P% %0 # -A~ V4 r -41 0 u 0A' * *S 44

i 0 *0 0 0 " A 0 .1 0 0 00. ca9 C.4 a r.-A C'4 V~ Ai 0 01 3 0 .94 -4 41 a o a. 0.- 4 5 '00V04

-4 Aj " -H -4 .r4 r4 4 5 06 4 $4 0 C30.W .44 3 w u 0A 004 404 0 01g 40 0 ~ rN 0-40

t - 44r6 ~4 r . 04 m 45 -4 64 600.45 I 0.0.404 V--. 4 04 u- --A 45 0i "44 r-a V4

4, " u 4 41 0* 0 ~ A 4V4 0 46. 41 -A0 va '- 4 .0 to In 60.6 &-4 a. H 40 a 0 v-" 0 0.04 a 0 * 4 0 .4 . -4 5'40

a v4 6r- 0~ 4 0 N *4 406.2 4 -(5 '- cov Wa-- u* -z.- 21- 1- u--~ 4 I N64 u-V4 u .

' 0 Cl (A4 to -'Efm

-H u4 66 6 63 9 4(1 -H 45 4 45 4 5U -A %I tU

26

0

0 P4

0r

010

010

001z00 w

0 000

4010j C: U

01* 011

C: 4J' r 4 4A 1-4 0 0 41

c4 0 0.0 0 0 03 0

-r4 0% fn ~ 0 N %D U

0 u ~ ~O 0 0 0A

01 V4V44 0a'V-4 0 0o

.0 W c0IV 416 0, 044w uM

00 &U I 0 0 0-A0"A .e r j t 1 u4 1 0 W c 41'

0 008 (A 0 0 .4(A .'v 00A A IAC01 44 0% C'J :3 :3 .e0 0C3

"4 H.4. -1 0 " w " 6

u 01ago.04 .. 01.0 0V4 L "41A 0

01 4 to I S PC, co Sol4 W ra, 00 ~4~ 0 4 0

U3 U

02

44

00 ca 6

IVL 414e.4c01 00

00

0 010 1440

$4 Lu V 4

0 440 90-x00 0 0 9Q

.0 44 r4V -4 0 .0 u w g

:P u 4 .1 $4 $ A c 0L

U- 4.5 :0 04.5 $ .V4 44 0 0 14 0

0 00

U 0 - .4 0

a ri So 0

144 OA N 44 C1 n4 4 4.50 C%0 U

1.4.045 020

Table 2. Seawater Contaminants Which Are Potentially Detrimental* to Reverse Osmosis Membranes

PretreatmentDetrimental Problem an Seawater NecessaryContaminant Detrimental Membrane Concentration For Normalor Factor Concentration Performance Ranite Occurrence

Silt,Sand Small Plugging, High Near YesFouling Surf Zone

Turbidity, I NTU Plugging, Variable YesSuspended FoulingSolids

Biological small Fouling, Variable YesMaterial Plugging

* (Algae)

Organics Fulvic & Hurnic Fouling, Variable Very(Color) Acids, Small Membrane Possibly

Swelling andDegradation;CA only

Oil and Small Fouling, Variable YesGrease Plugging

Iron Fe+2 A mg/I Membrane 0.002-0.9 mg/I Possibly

Fe*3 3'0.03 mg/I Fouling

Manganese )0.1 mg/I Membrane 0.001-0.05 NoFouling

PH <3 Membrane 74 No>9 Hydrolysis;

CA only

Temperature Varies with High Temperatureof Feed Percent Can Cause 60400F No

R!cgvery,0 Membrane<40 F, 40 F; CA Hydrolysis and0o, >113 0F; PA Compaction;

CA only

Silica 125 mg/I Membrane I mg/I NoFouling

CA a Cellulose acetate membranePA a Polyamide membrane

29

T!.....................

Table 3. Brackish Water Contaminants Which Are Potentially Detrimentalto Reverse Osmosis Membranes

PretreatmentDetrimental Problem an Brackish Water NecessaryContaminant Detrimental Membrane Concentration For Normalor Factor Concentration -Performance Range Occurrence

Fe*2 >0.0 mg/I oln

Maonane 01mI Membrane 0- 3.8 PossiblyFouling

SMlicans )125 mg/I Membrane 0-320 g/ PossiblyFouling

Calcium Slae Solubility FolnVariable YesCarbumSnate Index Precipitation

Stoniu ul ae calculation Scaling

Iron small Fouling, Iron Variable PossiblyBacteria Precipitation

On Membrane

Hydrogen small Fouling, Variable PossiblySulfide Sulfur No, except(Gas) Precipitation If present

On Membrane in ground-water

Swamp Humic and Fouling, Variable VeryOrganics Fulvic Acids, Membrane Possibly

small Swelling andDegradation;CA only

CA a Cellulose acetate membranePA a Polyamide memnbrane

30

Table 4. Specific Gray Water Contaminants Which Are PotentiallyDetrimental to Reverse Osmiosis Membranes

Potential ForCausing Problem

Concentration at ReportedConstituent mR/I Concentration

Aluminum hydroxide 0.5 LowAmmonium lauryl sulfate 3-14 LowCalcium carbonate 0.5 LowCastor oil 12-70 MediumCastor oil, sulf onated 3-16 MediumCoconut oil 5-16 MediumCoconut diethanol-amine 0.3-1 LowEpithelium cells 10 HighEthanol 10-50 LowEthoxylated alcohol 30 LowGlycerol 0.8-2 LowHair I High

* sopropyl alcohol 10-60?Kaolin, colloidal 3 HighKaolinite 70-90 HighLactic acid 3 LowMineral oil 0.2-9 MediumN,N -Diet hyl -m -toluamide 5-12?Oleic acid 10-30 MediumOlive oil, sulf onated 1-5Silica flour 50-110 HighSodium carbonate 240 HighSodiumn 4-diloro-2-phenylphenolate 0.3 LowSodium dodecylbenzenesulf onate 1-7?Sodium fluosilicate 35 MediumSodium ortho-phenylphenolate 0.3 LowSodium tripolyphosphate 40-45 HighSorbitol 0.4 LowStearic acid 6-17 HighTalc 20 HighTallow 1-21 HighTriethanolamine 0.5-2 LowTriethanolamide alkylbenzenesullonate 0.7-4 LowUltrawet 60L 3-14 LowVegetable oil 75 HighZinc Stearate I Medium

"?" Indicates insufficient data

31

Table 5. Classification of Feedwater Contaminants

Is ContaminantContaminant Water RemovalCategory Source Problem Necessary?

I. Solid lnorranicsa) Silt, sand, grit SW Plugging, Yesb) Heavy suspended

solids, turbidity SW Fouling Yesc) Suspended solids,

turbidity SW, BW, GW Yes

2. Solid Organicsa) Biological

Material (algae) SW, BSW Fouling Yesb) Oil and grease,

floatables SW, BSW Plugging Yesc) Emulsions SW, BSW Yesd) Iron bacteria GW Yes

3. Dissolved Organicsa) Swamp organics BWS, GW Fouling, membrane Nob) Organics BSW, SW swelling and No

degradation

4. Potential Precipitantsa) Iron, Manganese BW, SW, GW Fouling Possiblyb) Silica BW, GW No

5. Gasesa) Hydrogen Sulfide GW Fouling No

6. Low Solubility Saltsa) Barium Sulfate GW, BW Nob) Strontium Sulfate GW, BW Scaling Noc) Calcium Sulfate GW, BW, SW Nod) Calcium Carbonate GW, BW, SW No

GW = GroundwaterSW = SeawaterBW Brackish WaterBSW = Brackish Surface Water

413

m 32

- 14N

caa

N I.'

*~~~ 19 0 a S

o - o o..

1

o- :.~._: : :;§ N a

0

07

',s-I

,aNO NO::': 8::0 : :

*Z0 ac . ... I-

".- U-- 3*1*4

33 ~

0z <-J 2

<

Q).~~ La a~ 3

A a.

(1131j sOWunOlV) c

55O= U3~lUVd N N Nr4* N 0 0C

N OI.LV±OI! aw V __4 C 4 0 C

f7OdNOSLVWINO a NUJ - ~ 4

4.j A

0 MV MO. 0 0C 0 0 0 4 a 0Q

MOUVU~~tli ald2a N N (-4 4

04% 0 a 4 a9 C4 C4.

o) -L__ _ -

c_ _ _ 04 04 (4 N

0~ (Nnm N 0o N 000

(MAN NN N1 N - Ne.

r4)C

'5 ~ 0 0 0 0 0 0 0 N N N N(4 4 04 It I.

cc -

Q C4

.0

04U~l No"ow- CI 04 0 0 0 00O 0 0 0 0

-34

Table 8. Effect of Ultrasonics on the Flux Values ofMembrane Samples Fouled by Ferric Oxide

Sample Precleaning Ultrasonic Exposure Postcleaning aNo. Flux (gfd) (V;min) Flux (gfd) Change

IIA 17.1 50; 2 16.3 -0.8IIB 16.8 50; 2 17.4 +0.6lIC 17.7 none 18.0 +0.3lID 17.0 none 16.8 -0.2liA 19.8 50; 2 22.5 +2.7IIIB 18.5 50; 2 17.3 -1.2IIIC 15.7 50; 2 21.3 +5.6HID 17.3 50; 2 20.8 +3.5IIIE 16.8 none 18.2 +1.4IIIF 18.5 none 17.8 -0.7

+ increase in flux; - = decrease in flux.

Table 9. Ultrasonic Cleaning Efficiency MeasuredVisually for Membrane Samples Fouledby Calcium Salts

Ultrasonic Area CleanedSample No. Exposure (Z)

IA 50 V; 2 min. 22IB 50 V; 2 min. 26IC 50 V; 2 min. 45

Average 31

ID none 0

Ultrasonic CleaningEfficiency 31 - 0 = 31%

35

Table 10. Ultrasonic Removal Efficiency Based onVisual Inspection of Membrane SamplesFouled by Bentonite Clay

Sample No. Ultrasonic Exposure AreaClean Notes

1 100 V; 3 min 46 Minor membrane damage2 100 V; 3 min 60 Minor membrane damage4 150 V; 3 min 78 Major membrane damage

Mean 61

3 none 05 none 06 none 0

Ultrasonic Cleaning Efficiency 61- 0 =61%

Table 11. Ultrasonic Cleaning Efficiency Calculatedfrom Flux Values for Membrane SamplesFouled by Bentonite Clay

Sample Initial Precleaning Ultrasonic Exposure PostcleaningNo. (gfd max) (fouled;gfd ) (volts;min) (gfdf)

1 6.4 3.4 100; 3 8.52 5.9 3.1 100; 3 8.83 5.9 2.1 none 2.84 6.8 2.6 150; 3 9.75 7.5 3.4 none 3.36 6.8 3.4 none 2.6

Note:

(gfdf - gfdm)Percentage of flux recovered - gfd _ gfdm x 100

Percentage of flux recovered, Sample No. I - 170%Percentage of flux recovered, Sample No. 2 - 210%Percentage of flux recovered, Sample No. 4 - 170%

36

Table 12. Ultrasonic Cleaning of Single, Double,.- and Triple Layers of Membrane

Ultrasonic Exposure Area CleanDescription (volts; mn) (%)

Single membrane disk 50; 2 40(TFC 801 RO) 100; 2 90-95

One stack:Tricot, membrane, spacer,membrane, and Tricot 100; 2 5-7

Two stacks:

Tricot, membrane, spacer,membrane, Tricot, membrane,spacer, membrane, Tricot 100; 2 0-1

Table 13. Effects of Tricot and Vexar on UlcrasonicCleaning of RO Membrane

Ultrasonic Exposure Area CleanDescription (volts; mn) (%) Notes

Membrane (TFC 801 RO) 100; 2 70 Membranedamage

Membrane-modified spacer 100; 2 60

Membrane-standard Vexar

feed channel spacer 100; 2 50

Membrane-Tricot 100; 2 10

r"

-3

37 :

Table 14. Preliminary Flow-Surge Tests

Pcu Elapsed Minimum Final Flux Net: '" ProcedureTime Flux Flux Recovered A Flux-L', Tested

Tesed (hr:min) (gfd) (gfdf) (M) (M)

FS 25:15 10.5 12.3 32 -23FS 28:45 7.2 9.9 31 -38FS 44:15 5.5 7.9 23 -51

Average 7.7 10.0 29 -37

RD-FS 49.15 6.6 10.0 36 -38RD-FS 52:15 8.3 10.8 33 -33RD-FS 71:15 7.5 9.7 26 -39RD-FS 73:15 8.4 9.2 11 -43

Average 7.7 9.9 27 -38

BP-FS 74:15 8.5 9.8 17 -39BP-FS 76:15 8.5 11.2 36 -30BP-FS 92:45 6.5 9.1 27 -43

Average 7.8 10.0 26.7 -37

Note:

gfdmax - 16 (demonstrated before initiation of fouling feedtest, Figure 23)

(gfdf - gfdm )

Percentage of Flux Recovered - (gfdx gfdm) X 100

(gfdf - gfd x)Net Percentage of A Flux f ma x 100

gfd max

-. 38

v4 WJ

rJ44 * *. 2 r. C. -2 e-0 -l) 0% CLn e.4 %D 0n %D4 NA

00

V o

cd0 .4 N' CN -4 0l %DC2 0 0 " 0 0 . '-4Uf LM cn C14 cn1 C'4 IT C' CJr4 (n2 C14 %0

4) ok I I I I I I I I I I I I I1 -4 C4 4 -4 r, IT--. 0 -$ C. 4. 11 0%00 cc e r- -4 - C'J -7r %D '.0 IA - I

(A 00

4J

240 4-

4 4

4) ~44 IA 0 0 IA 0 0l 0l IA I IA 0 inCfl C-.

2-40

.0 44 ( 044 0"4 044044 0 44 04404404404&404.4 0:44@2 0 ca r 4-. -4 -- -4--. 141- -4 -. -4'-- 4 = -- -4 -- .- -4- @

00 2 @2 to wc o t e b 0 (so w2 0 to 00 W2 0 0 to ccb 06am 00 CI 00( 0 C-, 0I CmO0 U CI inIAL 00 0CIn 00 0Cin 00

w4 . $4 * o @2 4 A w4 w to 0 4 w 1 4 4 * 4 * w4 $* 4 -

4.0 0 -4 N -4 N 0 0 0 q 4 0 N N .N -INU -4 .3 .-4 N0.44 -4 -. 44 .4 - 44 - -4. -H -4. -A 4.4 .-"4 r4.44

W. 4.4 0 w401w44 44 4.4 @14.4 1 0 4 4q q 14 4@0 to r.:3 0 0 :

V1 w @m 14 'lA @ 2 W2 10 10 0 10 40 140 WO 1.0 140 40a) to 4) 0M0 2 (A @2A 0 0 @20 @2 @ 02

Z . 0 al 441 0 0 0 0 J40 0 j0 0@ 441 41 0 &j 444244, 10 w14 01401414 w4 14 C014 14 01 4014141414 N4Ui "q..0 U u u .40. .4.4 4 10..40.H0. me 0.

1 r4 '@2 a@ a@

@22 2 2 2 4 -4 '-4 en@2-

.4 . .4 ... .4- 4 CI f' ~ ..- 4 -4

(14

39

Table 16. Required Flow Rate for Mixed Media andTubular Fabric Filters Occupying theSame Superficial Working Area

Sand or Mixed-Media Filter Tubular Fabric Filter(gpm/ft2 ) (gpm/tube)*

1.0 0.028

1.5 0.042

2.0 0.056

3.0 0.056

4.0 0.111

5.0 0.139

6.0 0.167

10.0 0.278

15.0 0.417

*gpm/tube = gpM/ft2 .

40

Il w I

w A

- U

Iw

-Cz w

0.1.C

-o

LUU

CO1

w

00Wu

_j 4c

w4

-0 S - -

- ULTRASONIC TRAN SDUCERS

MEMBRANE

PRESSURE VESSEL

LAMINAR SUB-LAYER

Figur 2. IhW ewosf gd uinL aftomoc ituj% amdatm.

42

'If

41 4jE.04.4

- ILI

o *~'~43

CYCLONSTE CITE

Figure IM 4. ViblXpEtRtntstm.

N*I .I MIE

44"

.......................- -- -. V -. -. V S M- L .

4E4

4114J U

UUA 4.)

0 4-)

40041U

45I%:.7 UL

cii

4N 611

VoICU c;

cm a 0400

0 4.

uS e

(p46) xn-

C4

Al

I UL

let

c Qa

C

C 0

(U0

c.41I

o

qr.

c 00

4.48

0clJ

I-

W 4a

0 00

(o 41

'0 4J

0 -D

(p;6) 0nL

.~49

a 1%.

m

ID I

*5 - ~0U ii

Ii Ii4

In.'

I.5

I.3

.1

I*1

E~ii;i$ 50i ~ *1

N

b~SI~1

* ~iIlit

.0 IiiH!* 1.Ij

* 0

1% 10 N

(pjfi) xn~

50

Ii _u

.9 P

II 1'-t I

4

C"L

b.. o n cLnoU naL oL ooL n 0L

(pj6) xnL51~I

.1..ii%

CD

525

(4

CD

P6-

a CD D CD CD CD CD 0co Ln wr C0

U0133ar0

53:

--- fobl

23 .- --40 .-

0 04

I I' I II II

IL .

;.:. (PjB) xn LA

" 54

_ - -. '

iJ I

* 09 IdIht GD

* 0 1.& n n C6-;-C4 4(p;6) xn L

U 5-

-i

(%) P3~3f'

3i2I III

M VA

4-A-

1

La Lx eaL

-u 4..o

2U C.

CYCmb~

a.

C~CD0-0nciL

0C C4 U

(Ufi xn-

al56

. .. .

kn

HWa aY11"

IM.

*'-11

'57

-I-

.C

In u

0- -

00

*4 c

1 0

or C2G * 0 LCY CY V CY C(p46) xnL

C *58

Group III 290

80J

70

* 60

40230

20

10

0 - -LIINA 1118 IIIC HID IIIE IP

Membrane Sample

(Sesmple letam cespqond to those used in Tabe".

Figure 19. Percenage of ame cleaned a measured by visu a npection for membranes fouled by ferrcoxide (Sample letters corrstond to those used in Table I sad Figure 18).

59

.. - - -.

kn

c=C> -- w- =-D

U tn

4I 0 0 cm0

606

100

70-

6 0-

0

40

30-

20-

10-

0 IA 1B IC ID

Membrane Sample

Figure 2 1. Percentage of aea cleaned as measured by visual ispection for membranes foule by calciumsalts. (Sample letters correspond to those used in Table 9).

61

ucot permeate Carrier

Figure 22. Basic stack for ta in vertical ulrasonic cleaning appamus.

62

.' ..-'~~.. ..... ....-....-_ ...- ..... .--.--"-" ...- -...

..-.-. ... .-.. . . i : '-"- - - -/: " '- -. -: .": : -

- - - - -7-

- -n

* 4 0

Ur

.... a~An

C4 co cm

633

CV)

c now40

6 C-N

* -"

E at

Ik. - -7-

.4 0 4 400

V c ODC.

opt co4.lilt

C; 400

04

00 q

C co

0 004 4ca'.....................................

4~64

cJ

U UD* LL.4cii

U 0 U___ _ _ __

* 0*- --

~ = "111a

-0T 0-a r_ .0 x

- -0 - :5

* W 40

I -

o- -Doe-

)-

.1 0 0

x 140__

0--

* 0.G

(pj6) )"nLj

65

0 01

4-)

0l

0 ~ - -_ _ _c'J

CL___ %

0 4-

0) c. T-__ __ __ ____ __ __ __ CLI

'a 4a_ c__ __ _O__

j e4

C CLA

0 0 0-- _0 z -0IT

I C

LOLI0 AI 0OD

- a LA LO

4) LO0

C 4-- - . C-

4-)

CD 0 CM 2

0.11)~c *.EL0LJ - - - - - - - l0. -:)__

c'O .4'. CD. C\ - AL -DK0 0c CD 0\J 0tD.0 C\J 0) 0 0 r ci C(P16) xnL.J

66

-4 -

zz.3.

00

CO0

LUU

w-I>

I-0

cc LU

00

zolL

WJIMo U)

WUI.-

Lu *-H L-aC0>

C.)j

67

* ~ .

E

I-

66

-. sample point

Figure 29. Back view of the TF 2 experimental model.

69

04.

4- ai C

4-) E

. 0-

S(a

4- CL

700

4..

-,

.4.

I.

-:1

Figure 31. Fabric Type I split open to show dirt on internal fabric. ,7

71 -,

4* -- - * -- 4. r .. • .

Figure 32. Supports used inside the filter tubes for reverse flow.

72

-4J

CD-

EFFUEN

0 C

-

0- L

INLEN

LJu

-3

9j.oo 6.00 16.00 24.00 312. 00 l40.00TIME (MIN)

F~gur3 3. Flux dgine um.

73

*a I!

a IxPI

x 00.

xa

3-WL

474

* '.. * * .. in

"P us IL r

20 1-

'IL

Wc 0

IL-

916

A. c

u-|

75

.- 1|II

w -i

Appendix

GRAPHS OF PRESSURE VERSUS TIME FOR THE SEQUENCEOF FILTRATION/BACKWASHING EXPERIMENTS

SHOWN IN TABLE 15 ARE PRESENTED IN THIS APPENDIX

77

*. . -

° -. . . .

. . . . . .. . . . . . . . . . -

- - - - - - -. . . . . . ..

0. 'J

CL

co 12.00 24.00 36.0 4o a.o 00 b6.00: TIME (MI N)

;: ' Figure 36. Press= versus time for rowflow operation at 1.0 Ipm, Test no. 1.

a:

~~o°

, :

"' 12 . .00 ]2.O0 .00 3'G.OO I10.00 "

Figure 37. Presure veins tme for cross-fow opernios sil 4. gPm, Tes no. . ,

C)

a-

79 "

% --• .-. .• .- , , . I. ., .... . . .. . . . . . . ... .. .: . : " . • . : -. : . . . o (: - .:

* (.0

-occ'

U-,O

a-

'b. oo 5.0 b lb 0 20,00

crOl

• " TIME (MIN) i

: Figure 38. Preum versus ti for clse-end operation at 4.0 lpm, Ten: no. 3. .

i"U0

0%

C_.

0

b. 00 8 .0 10.00 52.00 32. OC 20.00

TIME (MINI

Figure 39. Pressure vrmus time for crou-flow operation at 4.0 in, Test no. 4.

so

C3-

'0'.00 6.00 12.00 16.00 24.00 30.00TIME (M IN)

Figure 40. Pressure versus time for crossflow operation at 1.0 gpm, Test no. S.

C CL1

CL

Cc

Tho i~: 20.00 30.00 4b..00 5 .00TIME (MIN)

Figure 41. Pressure versus time for crossflow operation at 4.0 gpm, Test no. 6.

CL

0

0

0

CL

0

L VU O 51.0i.o Ib 0 2.6 2.0

TIEoMN

CFigr 3 rsuevru iefrcosfo pmo t40gm etn.S

U-) 0 82

CL

IL-o

-

CD0

CL

ma-

0

0

c b116. 00 30.00 458.00 60.00 750.00TIME (MIN)

Figue 4. Pessure versus time for closedend operation at 1.0 pm, Test no..

08

0

C)

b. 00 ,6° .U 12,.00. i. 00 214 . 30 ,....

LUJ

ccma

0

9,.o. ,

Cy 100 00 12.00 18.00 214.00 30.00

TIME (MIN)

Figure 47. Pressure ves time for cloued-end opertion at 1.0 pm, Test no. 11.

d.4

Co

-. '-0

, c ='O 6"o"- - 6".o i' '

- .o 8oo 2.oo 0.0... TIE (MIN

,.,,o

°.U

"J . , .

Q

-0

.4.-A

-301-40

LU

Ccn

0

1500 30. 00 3b45 l.0 6000 75.00 oTIME (MIN)

Figur 48. Pressure versus dine for closed-end operstion at I.0 gpin, Tesn no. 13.

to.

0

0

CV

.0 b0 bK b o 6.0 70

TIE M NFiue4.Pesr esstm o eesefo pmina . pTs o 4

08

DISTRIBUTION LIST

ARMY Fat Engr. Letterkenny Army Depot, Chambersburg, PAAFB (AFIT/LDE), Wright Patterson OH; CESCH, Wright-Patterson; HO Tactical Air CmdIDEMM (Schmidt)

Langley, VA; MACIDET, Scott, IL; SAMSOIMNND, Norton AFB CA; Stinfo Library, Offutt NEAFESC DEB, Tyndall. Fl.ARMY ARRADCOM, Dover, NJ; BMDSC.RE (H. McClellan) Huntsvific AL; Contracts - Facs Engr

Directorate, Fort Ord, CA; DAEN-CWE-M, Washington DC; DAEN-MPE-D Washington DC;DAEN-MPU. Washington DC; Engr District (Memphis) Library, Memphis TN; Tech. Ref. Div., FortHuachuca, AZ

ARMY -CERL Library, Champaign ILARMY COASTAL ENGR RSCH CEN Fort Belvoir VAARMY COE Philadelphia Dist. (LIBRARY) Philadelphia, PAARMY CORPS OF ENGINEERS MRD-Eng. Div., Omaha NE; Seattle Dist. Library, Seattle WAARMY CRREL Library, Hanover NHARMY DARCOM Code DRCMM-CS Alexandria VAARMY ENG DIV HNDED-CS, Huntsville AL. HNDED-FD, Huntsville, ALARMY ENO WATERWAYS EXP STA Library, Vicksburg MSARMY ENGR DIST. Library, Portland ORARMY ENVIRON. HYGIENE AGCY Dir Env Qual Aberdeen Proving Ground MD, HSE-EW Water Qual

Eng Div Aberdeen Proy Gmud MD; HSE-RP-HG/Pest Coord, Arberdeen Proving Ground, MD; Librarian,Aberdeen Proving Ground MD

ARMY MATERIALS & MECHANICS RESEARCH CENTER Dr. Lenoe, Watertown MAARMY TRANSPORTATION SCHOOL Code ATSPO CD-TE Fort Eustis, VABUMED Security Offr, Washington DCBUREAU OF RECLAMATION Code 1512 (C. Selandev) Denver COCINCLANT CIV ENGR SUPP PLANS OFFR NORFOLK, VACINCPAC Fac Engrng Div (J44) Makalape, HICNAVRES Code 13 (Dir. Facilites) New Orleans, LACNM Code MAT-04, Washington, DC; NMAT - 044, Washington DCCNO Code NOP-964, Washington DCCOMCBPAC Operations Off, Makalapa HICOMFLEACT, OKINAWA PWO, Kadena, OkinawaCOMNAVBEACHPHIBREFTRAGRU ONE San Diego CA-COMNAVMARIANAS Code N4, GuamCOMOCEANSYSLANT PW-FAC MGMNT Off Norfolk, VADEFENSE INTELLIGENCE AGENCY DB-4CI Washington DCDEFFUELSUPPCEN DFSC-OWE (Term Engrng) Alexandria, VADLSIE Army Logistics Mgt Center, Fort Lee, VADOD Staff Spec. Chem. Tech. Washington DCD71C Defense Technical Info Ctr/Alexandria, VADThSRDC Code 42, Bethesda MDDTNSRDC Code 522 (Library), Annapolis MDENVIRONMENTAL PROTECTION AGENCY Reg. III Library, Philadelphia PA; Reg. ViII, 8M-ASL,

4 Denver CO. Reg. X Lab. (M/S 541), Seattle WAFLTCOMBATTRACENLANT PWO, Virginia Bch VAFMFLANT CEC Offr, Norfolk VAGSA Assist Comm Des & Cnst (FAIA) D R Dibner Washington, DC ;Fed. Sup. Serv. (FMBP). Washington

DC; Off of Des & Const-PCDP (D Eakin) Washington, DCHCU ONE CO, Bishops Point, HIKWAJALEIN MISRAN BMDSC-RKL-CLIBRARY OF CONGRESS Washington, DC (Sciences & Tech Div)MARINE CORPS BASE Code 406, Camp Lejeune. NC; M & R Division, Camn Lejeune NC; Maint Off Camp

Pendleton, CA; PWD - Maint. Control Div. Camp Butler, Kawasaki, Japan; r'WO Camp Lejeune NC;PWO, Camp S. D. Butler, Kawasaki Japan

MARINE CORPS IJOS Code LFF-2, Washington DCMCAS Facil. Engr. Div. Cherry Point NC; CO, Kaneohe Bay HI; Code S4, Quantico VA; Facs Maint Dept-

Operations Div, Cherry Point; PWD - Utilities Div, Iwakuni, Japan; PWD, Dir. Maint. Control Div.,Iwakuni Japan; PWO, Iwakuni, Japan; PWO, Yuma AZ

MCDEC M&L Div Quantico VA; NSAP REP, Quantico VAMCLB 8520, Barstow CAMILITARY SEALIFT COMMAND Washington DCNAF PWD - Engr Div, Atsugi, Japan; PWO, Atsugi JapanNALF OINC, San Diego, CA

* -... 86

NARF Code 100, Cherry Point, NC; Code 612, Jax, FL; Code 640, Pensacola FLNAS CO, Guantanamo Bay Cuba; Code 114, Alameda CA; Code 183 (Fac. Plan BR MGR); Code 197,

Jacksonville FL; Code 18700, Brunswick ME; Code 18U (ENS P.J. Hickey), Corpus Chuisti TX; Code 70,Atlanta, Marietta GA; Code BE, Patuxent Riv., MD; Dir of Engrng, PWD. COWpu Christi, TX; Dir. Util.Div., Bermuda; PW (3. Maguire), Corpus Christi TX; PWD - Engr Div Dir, Millington, TN; PWD - EngrDiv, Otmo, Cuba; PWD - Engr Div, Oak Harbor, WA; PWD Maint. Div., New Orleans, Belle Chasse LA;

* PWD, Willow Grove PA; PWO Belle Chasse, LA; PWO Chase Field Beeville, TX; PWO Key West FL;PWO Lakehurst, NJ; PWO Sigonella Sicily; PWO Whiting Fid, Milton FL; SCE, Cubi Point, R.P; PWO,Dallas TX; PWO, Glenview IL; PWO, Kingsville TX; PWO, Millington TN

NAVFACENGCOM CONTRACT'S ROICC Key West FLNAS SC! Norfolk, VANARF SCE Norfolk, VA

NAS SCE, Barbers Point HI* NATL BUREAU OF STANDARDS R Chung Washington, DC

NATL RESEARCH COUNCIL Naval Studies Board, Washington DCNAVAEROSPREGMEDCEN SCE, Pensacola FLNAVAIRDEVCEN Chmielewski, Warminster, PA; PWD. Engr Div Mgr, Warminster, PA

* NAVAIRPROPTESTCEN CO, Trenton, NJNAVAIRTEST7CEN PATUXENT RIVER PWD (F. McGrath), Patuxent Riv.,MDNAVAVIONICFAC PWD Deputy Dir. D/701, Indianapolis, INNAVCHAPGRU Operations Officer, Code 30 Williamsburg, VANAVCOASTYSCEN Code 715 (3 Quirk) Panama City, FL; Code 772 (C B Koesy) Panama City FL; Library

Panama City, FL; PWO Panama City, FLNAVCOMMAREAMSTRSTA PWO, Norfolk VA; SCE Unit I Naples Italy; SCE, Wahiawa HINAVCOMMSTA Code 401 Nea Makri, Greece; PWD - Maint Control Div, Diego Gardia Is.; PWO, Exmnouth,

Australia; SCE, Balbo., CZNAVCONSTRACEN Code 0OU 15, Port Hueneme CA

* NAVEDTRAPRODEVCEN Technical Library, Pensacola, FLNAVEDUTRACEN Engr Dept (Code 42) Newport, RINAVEODTECHCEN Code 605, Indian Head MD

* NAVFAC PWO, Centerville Bch, Ferndale CA; PWO, Point Sur, Big Sur CANAVFACENGCOM Code 03T (Essoglou) Alexandria, VA; Code 043 Alexandria, VA; Code 044 Alexandria,

VA; Code 044B) Alexandria, VA;, Code 0451 (P W Brewer) Alexandria, Va;, Code 04548 Alexandlria, Va; p

Code 04AI Alexandria, VA; Code 051A Alexandria, VA; Code 06, Alexandria VA; Code 1113, Alexandria,VA

NAVFACENGCOM CONTRACT'S ROICC, Yap; ROICC Code 495 Portsmouth VANAYFACENOCOM code OfT Alexandria, VANAVFACENGCOM - CHES DIV. Code 405 Wash, DCNAVFAC!NGCOM - LANT DIV. Code 403, Norfolk, VA; Code 405 Civil Engr BR Norfolk VA; Eur. BR

Deputy Dir, Naples Italy; RDT&ELO 102A, Norfolk, VANAVFACENGCOM - NORTH DIV. (Boretsky) Philadelphia, PA; CO; Code 04 Philadelphia, PA; Code 09P

Philadelphia PA; Code 1028, RDT&ELO, Philadelphia PA; Code III Philadelphia, PA; Code 114 (A.Rhoads); ROICC, Contracts, Crane IN

NAVFACENGCOM - PAC DIV. CODE 09P PEARL HARBOR HI; Code 2011 Pearl Harbor, HI; Code 402,RDT&E, Pearl Harbor HI; Commander, Pearl Harbor, HI

NAVFACENGCOM - SOUTH DIV. Code 90, RDT&ELO, Charleston SCNAVFACENGCOM - WEST DIV. AROICC, Contracts, Twentynine Palms CA; Code 04B San Bruno, CA; _

Code 114C. San Diego CA-. 09P/20 San Bruno, CA; RDT&ELO Code 2011 San Bruno. CANAVFACENGCOM CONTRACTS AROICC, Point Mugu CA; AROICC, Quantico, VA; Dir, Eng. Div.,

Exmouth, Australia; Eng Div dir, Southwest Pac, Manila, P1;, 01CC, Southwest Pac, Manila, P1;OICC-ROICC, NAS Oceana, Virginia Beach, VA; OICCIROICC, Balboa Panama Canal; ROICC AF Guam;ROICC, Keflavik, Iceland;, ROICC, NAS, Corpus Christi, TX; ROICC, Pacific, San Bruno CA;

ROICC-OICC-SPA, Norfolk, VANAVFORCARIB Commander (N42), Puerto RicoNAVMAG PWD - Engr Div, GuamNAVOCEANO Library Bay St. Louis, MSNAVORDSTA PWO, Louisville KYNAVPETOFF Code 30. Alexandria VANAVPETRES Director, Washington DCNAVPGSCOL E. Thornton, Monterey CANAVPHIBASE CO, ACE 2 Norfolk, VA; Code S3T, Norfolk VA; SCE Coronado, SD,CANAVRADRECFAC PWO, Kami Seya Japan_NAVREGMEDCEN Code 29, Env. Health Serv. (Al Bryson) San Diego, CA; SC!, Newport, RINAVREGMEDCEN PWO. Okinawa, JapanNAVREGMEDCEN SC!; SC! San Diego, CA; SC!, Camp Pendleton CA; SCE, Oakland CANAVREGMEDCEN SCE, Yokosuka, Japan

87

NAVREGMEDCLINIC A. Watanabe, Pear Harbor, HINAVSCOLCECOFF C35 Port Hueneme, CA; CO. Code C44A Port Hueneme, CANAVSCSOL PWO, Athens GANAVSEASYSCOM SEA 04E (L Ken) Washington, DCNAVSECORUACT Facil. Off., Galeta Is. Panama Canal; PWO, Adak AK; PWO, Edzell Scotland; PWO,

Puerto RicoNAVSHIPYD Bremerton, WA (Carr Inlet Acoustic Range); Code 202.5 (library) Puget Sound, Bremerton

WA; Code 380, Portsmouth, VA; Code 400. Puget Sound; Code 410, Mare Is., Vallejo CA; Code 440Portsmouth NH; Code 440, Norfolk; Code 440, Puget Sound, Bremerton WA; L.D. Vivian; Library,Portsmouth NH; PW Dept, Long Beach, CA; FWD (Code 420) Dir Portsmouth, VA; PWD (Code 450-HD)Portsmouth, VA; PWO, Bremerton, WA; PWO, Mare Is.- PWO, Puget Sound; SCE, Pearl Harbor HI; TechLibrary, Vallejo, CA

NAVSTA CO Roosevelt Roads P.R. Puerto Rico; CO. Brooklyn NY; Code 4, 12 Marine Corps Dist, TreasureIs., San Francisco CA; Dir Engr Div, PWD, Mayport FL; Engr. Dir., Rota Spain; Long Beach, CA; Maint.Coat. Div., Guantanamo Bay Cuba; Maint. Div. Dir/Code 531, Rodman Panama Canal; PWD - Engr Dept,Adak, AK; PWD - Engr Div, Midway Is.; PWO, Keflavik Iceland; PWO, Mayport FL; SCE, Guam; SCE,Subic Bay, R.P.; Security Offr, San Francisco, CA

NAVSUBASE SCE, Pearl Harbor HINAVSUPPFAC PWD - Maint. Control Div, Thurmont, MDNAVTECHTRACEN SCE, Pensacola FLNAVWPNCEN Code 2636 China Lake; Code 3803 China Lake, CA; PWO (Code 266) China Lake, CANAVWPNSTA Code 092, Colts Neck NJ; Maint. Control Dir., Yorktown VANAVWPNSTA PW Office Yorktown, VANAVWPNSTA PWD - Maint Control Div, Charleston, SC; PWD - Maint. Control Div., Concord, CA; PWD -

Supr Gen Engr, Seal Beach, CA; PWO, Charleston, SC; PWO, Seal Beach CANAVWPNSUPPCEN Code 09 Crane INNCIC Coast. Elec. School, Port Hueneme, CA

.. NCBC Code 10 Davisville, RI; Code 15, Port Hueneme CA; Code 155, Port Hueneme CA; Code 156, Port*Hueneme, CA; Code 25111 Port Hueneme, CA; Code 400, Gulfport MS; Code 430 (PW Engrng) Gulfport,

MS; NEESA Code 252 (P Winters) Port Hueneme, CA; PWO (Code 80) Port Hueneme, CA; PWO,Guport, MS

NCR 20, Code R70; 20, CommanderNMCB 74, CO; FIVE, Operations Dept; Forty, CO; THREE, Operations Off.NOAA ibrary Rockville, MDNORDA Code 410 Bay St. Louis, MS; Code 440 (Ocean Rsch Off) Bay St. Louis MSNSC Code 54.1 Norfolk, VANSD SCE, Subic Bay, R.P.NSWSES Code 0150 Port Hueneme, CANTC OICC, CBU-401, Great Lakes ILNUCLEAR REGULATORY COMMISSION T.C. Johnson, Washington, DCNUSC Code 131 New London, CT; Code EA123 (R.S. Munn), New London CT; Code SB 331 (Brown),

Newport RIONK Code 70F Arlington VAPHIBCB 1 P&E, San Diego, CA; 1, CO San Diego, CA; 1, CSWC D Wellington, San Diego, CAPMTC Pat. Counsel, Point Mugu CA

.•PWC CO Norfolk, VA; CO, (Code 10), Oakland, CA; CO, Great Lakes IL; CO, Pearl Harbor HI; Code 10,Great Lakes, IL; Code 105 Oakland, CA; Code 120, Oakland CA; Code 120C, (Library) San Diego, CA;Code 128, Guam; Code 154, Great Lakes, IL; Code 200, Great Lakes IL; Code 30V, Norfolk, VA; Code400, Great Lakes, IL; Code 400, Oakland, CA; Code 400, Pearl Harbor, HI; Code 400, San Diego, CA;Code 420, Oakland, CA; Code 424, Norfolk, VA; Code 505A Oakland, CA; Code 600, Great Lakes, IL;Code 610, San Diego Ca; Code 700, Great Lakes, IL; Code 700, San Diego, CA; Util Dept (R Pascua) PearlHarbor, HI; Utilities Officer, Guam

SUPANX PWO, Williamsburg VATVA Solar Group, Arnold, Knoxville, TNUS DEPT OF HEALTH, ED., & WELFARE Food & Drug Admin, (A. Story), Dauphin Is. AL

* US GEOLOGICAL SURVEY Off. Marine Geology, Piteleki. Reston VAUS NAVAL FORCES Korea (ENJ-P&O)USAF REGIONAL HOSPITAL Fairchild AFB, WAUSCG (Smith), Washington, DCUSDA Forest Service Reg 3 (R. Brown) Albuquerque, NM; Forest Service, Bowers, Atlanta, GA; Forest

Service, San Dimas, CAUSNA Civil Engr Dept (R. Erchyl) Annapolis MD; ENGRNG Div. PWD, Annapolis MD; Energy-Environ

Study Grp, Annapolis, MD; Environ. Prot. R&D Prog. (J. Williams), Annapolis MD; NAVSYSENGRDept. Annapolis, MD; PWO Annapnlis MD

BROOKHAVEN NATL LAB M. Steinberg, Upton NYGEORGIA INSTITUTE OF TECHNOLOGY (LT R. Johnson) Atlanta, GA

~88

.., ... '. -. . / ... . ' /. .... ...... .. . .. ' >". ... . . - ., - . . . . . .

3m8

I

II

4

a1