AN ERVESHGAHON OF THE PURIFICAECN OF SYNTHETIC BRINE WITH REVERSE OSMOS§S “5515315 fer the Dams m" 321 S. amaze-rm 3m??? ‘vmmm i; mm mm ‘ 29m A-n— ‘ .. “mi-w LIBRARY Michigan Stilt? University w ABSTRACT AN INVESTIGATION OF THE PURIFICATION OF SYNTHETIC BRINE WITH REVERSE OSMOSIS By J. Lanny Tucker A study was made of the use of reverse osmosis to re- move inorganic ions from tap water and a synthetic brine solution. The equipment used included a pretreatment filter, an in—line mixing tank, an acid feed pump, and a high pres- sure pump. The system had appropriate pH, temperature, pressure, and flow rate instrumentation and controls. Water purification proceeded with a hollow fine fiber R0 module manufactured by Dow Chemical Company. The module fibers were made of cellulose acetate and have an outside diameter of 50 microns. The ion rejection of so; mined over a period of time. Water recovery and water flux were calculated. A unique feature of the system was the ability to recycle waste water back into the feed. The equipment was first tested with tap water. Later experimen— tation was designed to find the ion rejection of Na+ and so;2 in a synthetic brine that had a concentration of over 8700 ppm of inorganics. Analysis was made by injecting radioisotopes of the ion studied into the feed stream. This injection occurred with the hydrochloric acid used to control the pH between the limits of 5.5 and 6.0. Mixing of the tracers in the feed was assured in a 7 gallon mixing tank. Radioactivity was detected with liquid scintillation techniques. Data were obtained by taking samples of feed, per- meate, and concentrate streams. The resulting net counts were used to calculate ion rejection. In the tap water experiments, rejection of all three ions was found to reach near 95% at steady state. A lower 2, Ca++, and Na+ were deter— water recovery(percentage of feed recovered as permeate) increased the ion rejection, while an increase in the amount of waste recycled lowered the product water qual- ity. These results agree with those found in the liter- ature for brackish waters(l4). In addition, the divalent ions, 8012 and Ca"+ were rejected more easily than the monovalent Na+. Experimentation with the synthetic brine produced consistent results with Na+. Rejection was near 90% at steady state with a recovery of 70%. Product water qual- ity declined when water recovery was increased. Results from experiments with the sulfate ion were not consistent enough to draw conclusions. The reverse osmosis system behaved as if the module contained some carbonates and sulfates which lowered product quality. The use of reverse osmosis in the Lansing area to purify brackish water appears to be feasible from the standpoint of the technique. Further studies using Marshall formation brine should be performed to investi- gate the economic feasibility. AN INVESTIGATION OF THE PURIFICATION OF SYNTHETIC BRINE WITH REVERSE OSMOSIS By M“ J: Lanny Tucker A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemical Engineering 1972 To Jesus Christ iii ACKN WLEDGEENTS The guidance and thoughtful advice of Dr. Bruce W. Wilkinson was gratefully received by the author during the course of this study. His helpful suggestions and provo- king questions are sincerely appreciated. Appreciation is also extended to Dr. M.H. Chetrick for finding available financial support for the Masters program which included this research. The author expresses gratitude to Mr. Gerry T. West~ brook of Dow Chemical Company for his help, and for his arrangements for use of the Dow Reverse Osmosis unit. Mr. Don Childs merits special thanks for fabricating parts and making the equipment Operable. iv TABLE OF CONTENTS IntrOductionOOO000.00.00.00.0COOOOOOOOOOCOOOOOCOOOO 1 TheoryOOI....0O0.0......0.0'.OOOOOOOOOOOOOOOOOOOOOC 5 Reverse Osmosis Operation.................... 5 Membrane Failure............................. 6 Membrane Transport mechanisms................ 8 Membrane Configurations...................... 9 Reverse Osmosis System Parameters............13 Experimental Equipment.............................17 Experimental IiiethOdso0.0.0.000.00.00.00..000000000022 Safety Precautions...........................24 Analytical rflethOdSOOOCOOOOO0.0.0.0000...0.00.00.00.26 DataOOOOOOOOOOOO0.00.0000...0.0.0.000...0.0.0.0....28 ResultSO00....OOOOOOOQOOOOOOOOOOOOOO00.000.00.0000035 Results With Tap Water.......................35 Results With Synthetic Brine.................39 Analij-s Of ResultSOOOO0.0.00....OOOOOOOOOOOOOOOOOOSO Tap war-hero...COO...00.00.000.00...0.00.00.00.50 Synthetic Brine..............................52 Conclusions and Recommendations....................54 Basic Observations...........................54 Suggested Research...........................55 Suggested Experimental Improvements..........55 Bibliogr‘aphy0.00.00.000.000.0.00.00.00.000.00.00.0056 Appendix A...OI...000000....OOOOCOOOOOOOOOOOOOOOOO059 Scintillation Spectrometry...................59 Neutron Activation and Radioactive Decay.....63 Appendix B ooocccoococooooc000000000.0000000000000065 Appendix C cocoocooooooooocooocoooococo00000000000067 AppendixDOOOOOOOOOO.O0...0.0.0.00.00.00.00000000074‘ vi Table 1. Table 2. Table 3. Table 4. Table 5. Table 6. Table 7. Table 8. Table 9. Table 10. LIST.OF TABLES Dow Hollow Fiber Module Specifications...................21 Experimental Data Obtained for so“2 and Ca++ Ions Using Tap Water as Feed Stream.........29 Experimental Data Obtained for Na+ Ion Using Tap Water as Feed Stream....o...oo.....o......30 Experimental Data Obtained for Na+ Ion Using Synthetic Brine as Feed Stream.............3l Experimengal Data Obtained for SO- Ion Using Synthetic Brine as Feed Stream.............32 Activities of 9 ml Samples of Mixing Tank Flush................33 Data Obtained from System Flushing to Determine Residual Activity................34 Tap Vlater A.nalyseSOO000000.00000000000065 Synthetic Brine Analysis...............65 Chemical Makeup of Synthetic Brine...00.00.0.00.0000000000000066 vii Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 9. 10. ll. 12. LIST OF FIGURES Basic Reverse Osmosis Plant Process Diagram...............2 Basic Principles of Osmosis and Reverse Osmosis...........5 Plate and Frame Module configwatiODOOOOO0.00.00.00.10 Tubular Module Configuration........10 Spiral Wound Module Configuration................ll Dow Hollow Fine Fiber R0 RIOduleOOOOOOOOOOO0.0.0.00000012 Flow Diagram of Reverse 081110818 systemOOOOOOOOOOOOOOOJ-S Front View of Dow Reverse 081110313 unitOOOOOOOOOOOOOOOOOlg Rear View of Dow Reverse osmOSiS UnitOOOOOOOOOOOOOOOOOZO Plot of % Ion_§ejection vs. Time for SO Ion at 70% Water RecoveTy Using Tap Water as Feed Stream...............37 Plot of % Ion_§ejection vs. Time for SO Ion at 50% Water RecoveSy Using Tap Water as Feed Stream...............38 Plot of % Ion Rejection vs. %2 Water Recovery for SO4 Ion after 120 Minutes Using Tap Water as Feed StreamOOOOOOOOOOOOO0.00.00.004'0 viii Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure l3. 14. 15. 16. 17. 18. 19. 20. 21. 22. Plot Plot Plot Plot Plot Plot Plot Plot of % Ion Rejection vs. Time for Ca++ Ion with 60% Recycle Using Tap Water as Feed Stream....................4l of % Ion Rejection vs. Time for Ca++ Ion with No Recycle Using Tap Water as Feed StreamOCOCOOOOOCOCOOC0.0000000042 of % Ion Rejection vs. Time for Na+ Ion with 60% Recycle Using Tap Water as Feed StreamOOOOOOOC.0000000000043 of % Ion Rejection vs. Time for Na+ Ion with No Recycle Using Tap Water as Feed Stream-00.0.0...0.0.00000004‘4 of % Ion Rejection vs. Time for Na+ Ion at 80% Recycle Using Synthetic Brine as Feed Stream-00......000......46 of % Ion Rejection vs. Time for Na+ Ion with No Recycle at 70% Recovery in Synthetic Brine...0.0...COCOOCOOCOOOOOQOC47 of % Ion Rejection vs. % Water Recovery for Na+ Ion after 60 Minutes Using Synthetic Brine as Feed StreamOOOOOCOOOOOOOO00.0.0....048 of % Ion Rejection vs. Time (discontinuous) for SO Ion Using Synthetic Brine gs Feed Stream....................49 Block Diagram of Basic Scintil- lation Counting System.........59 Packard Instrument Co. Scheme of Liquid Scintillation system-0.00.00.00.00.00.00.000.062 ix Figure 23. Figure 24. Figure 25. Activity v25 % Gain for S35 and Ca Scintillation DetectionOOOOOOOOO000.000.00.078 Activity vs. % Gain for Na24 Scintillation Detection.......79 Plot of Permeate Flux(Relative to 2500) vs. Operating Temperature for DDW’HFF P'iOduleI.0.0.00.00000000000000080 INTRODUCTION Tremendous interest has been generated during the past fifteen years for reverse osmosis as a promising method for economic recovery of water from saline solu- tions. Using this technique, desalination is achieved by forcing a salt solution under pressure past a semi- permeable membrane which passes water more readily than organic material or inorganic ions. A flowsheet of a typical process system is shown in Figure l. The method differs from electrodialysis in that the driving force is pressure instead of electrical potential. . Reverse osmosis has several distinct advantages. The method is simple in concept; Operation is at ambient Ctemperature; and membranes are being continually devel- Oped which give successively better rejections and higher fluxes at lower pressures. In fact, the technology has moved so fast that current desalting plant operations are feasible and available for municipal and industrial use in plants with capacities up to 10 million gallons per day(M GPD). Desalting in large quantities of 50M GPD or more can be practically considered for Operation in the late 1970's or early l980's(20). It appears that the first proposal that salt rejec- tion by membranes might be useful in desalination was made by C.E. Reid at the University of Florida to the Office of Saline Water in 1953(16). Research was begun by Loeb and Sourirajan at UCLA to develop a cellulose acetate membrane with a high flux. This work was con- tinued by General Dynamics and Aerojet General up to the present day. Today, research in reverse osmosis is carried out by many universities and corporations. Much of this work is sponsored by the Office of Saline Water 1 Emammfln mmOOOHm pfide Om OHmmm .H ehsmflm Made OmMHOpm~IDKMMHFIJ.. \i Hepm3 godeohm Esehpm podwohm —1» L. r J mESm mmooohm .lll. OhSmmohm QmHm ” mfidm Hmpmzpoom moassoz g . .lbr.’ Lb. mESm fiospohm soapacs< Hooasoso soap usapaaa » sees mscaom _ madam soap ussaaoaso JVpll Omhsfiomfin hrwm , QESA onflhm pomfiem osspsH 3 in the U.S. Department of the Interior. It is carried out in all parts of the world, and is subject to compar- ison with other desalination techniques such as freezing, flash distillation, electrodialysis, exchange diffusion, and piezodialysis. Much attention has been paid by researchers in re- verse osmosis to Obtaining fresh water from the sea. But of more interest to inland experimenters is the economic feasibility of using the method to purify brackish waters, industrial wastes, and municipal drain—Off and sewage. Power requirements and other economic factors point out that reverse osmosis is advantageous over other processes such as distillation and electrodialysis in purifying brackish waters. Reverse osmosis removes the organics or ions from the fresh water while the other methods remove the water from the contaminants and require the same energy for every feed stream(l4). The Federal government has sponsered development-of a 250,000 GPD test bed for brackish water(20). Of primary emphasis to its users is development of membranes with capacities Of 10 or 50 times the present cellulose acetate ones. Recent research has accomplished the following(l9): 1. Development of blend cellulose acetate membranes which give higher production rates for brackish water. 2. DevelOpment of the novel composite concept(ultra- thin film barrier on a micrOporous membrane backing or support). 3. Discovery of efficient membranes of cellulose acetate butyrate. 4. DeveIOpment of highly specialized films, tubes, and hollow fibers for purification. 5. Preparation of cellulose acetate fibers by the unique thermal precipitation method. 6. Progress in development of technology for in plant removal and membrane replacement. 4 It has been known for many years that below the fresh water table in the central region of Michigan is a vast deposit of brackish water. This deposit of water is called the Marshall formation and can be found at a depth of about 600 feet(l8). Late 19th Century drillers had hoped the salt content would be high enough so that comm- ercial sale of the minerals would be profitable. A U.S. Geological Survey of 1900 showed the concentration to be about 8768 ppm(l8). This includes calcium, sodium, chlo— ride, sulfate, and bicarbonate as the principal ions. The synthetic brine described in Appendix B closely re- sembles the composition of the Marshall formation. Today, thought is being given to this brackish source as potential municipal water. Reverse osmosis is looked upon as one of the best means of purifying the Marshall formation. The purpose Of the present study was to become ac- quainted with the reverse osmosis process and equipment, to determine particular ion rejection rates both from tap water and a synthetic brine solution, and to Obtain data on the membrane fluxes of the equipment. The equipment was provided by Dow Chemical Company with the understanding that they would be allowed any and all information derived from its use. Analysis was accom- plished by using radioactive tracers injected into the feed stream. NO report in the literature has been made as to use of this technique in reverse osmosis research. However, it is being currently used in experimentation sponsered by the Office of Saline Water of the U.S. De- partment of the Interior at Clarkson College. THEORY Reverse Osmosis Operation Osmosis, as well as reverse osmosis, depends upon the existence of a membrane that is selective in the sense that some components of a solution can pass through but not the others (semi-permeability). Both processes can be simply illustrated as in figure 2. ‘ EMI Earring ' Sahne ? Fresh If» FVCSIN H10 C “7.0 a we? ”2.0 (3.0115) ’3 “no a i, “to I ‘- (II/ H30 / H‘o Osmosis Osmotic Equilibrium gfaesk 1.? H10 Figure 2. Basic Principles ; of Osmosis and m Reverse Osmosis “go—6* / Reverse Osmosis Under normal conditions, water will flow from the fresh container to the concentrated solution. The drive ing potential is called the Osmotic pressureCAflj. Actu- al flow of fluid is related to the chemical potential of the solution, which is a function of the solution pres- sure, temperature, and the number and types of molecules in solution. When an external pressure(equal to the osmotic pres- sure) is applied to the concentrated salt solution, the flow of fluid is in equilibrium. As this external pres- sure is increased, water is forced from the concentrated 6 solution into the fresh water container, leaving the ions behind. This water flux can be described by: J1 = K1(AP am) (1) or, J1 = KlPeff (2) where, Peff = the effective membrane driving pressure AP’ = pressure difference between the feed stream and the product stream K1 = membrane permeability constant(GPD/ft2-psi) Jl = product water flux(GPD/ft2). The movement of ions across the membrane, though not desirable, cannot be entirely stOpped. This flux can be approximated by: J2 = mow - op) (3) where, J2 = salt transfer flux(lb/hr) B = salt permeation constant C = feed stream salinity measured at membrane w wall CD = product salinity. Equation 1 indicates the significance of high fluid pressure on the water production rate. Acting in Opposi— tion to the pressurized fluid is the osmotic pressure of the saline solution. Osmotic pressures associated with typical brackish water feeds are 30-150 psi, whereas sea water may exhibit pressures as high as 450-600 psi. Therefore, reasonable water flux through current high salt rejecting cellulose acetate membranes requires pres- sures of 600-800 psi for brackish water, and up to 1500 psi for sea water(l4). Membrane Failure Ideally, reverse Osmosis membranes would pass all of the water forced through the system. But this objective 7 is difficult to obtain in practice. A decrease in water flux occurs for the following reasons: 1. Membrane fouling by scale, contaminants, etc. 2. Membrane compaction and compression—~related to the effects associated with the sustained high pressure fluid crushing the porous structure of the membrane. 3. Irreversible hydrolytic deterioration of the mem- brane material-- related to the hydrolysis of the cellu- lose acetate to cellulose. Fouling is reduced by proper pretreatment, both fil— tering and chemical, to remove organics, large sediments, iron, and manganese. Organics and sediments of silica are best removed by a system of filters in front of the re- verse osmosis system. Iron and manganese removal is not quite as simple. It requires passing the water through zeolite sands of KMnO4 where Fe and Mn are oxidized so that insoluble oxides are formed. These are subsequently filtered out(25). Compaction is the long term degradation Of the mem- brane. In—situ methods proposed to counteract this effect and to restore membranes to their original performance have largely failed to date(l4). It appears that compac— tion is best reduced through proper design and manufacture of the membranes. Hydrolytic degradation of cellulose acetate to cellu- lose is reduced only by controlling the pH of the feed within certain limits. The ability of the membranes to reject salts also decreases with time. The major factors appear to be(l4): l. Membrane compaction-- the effect of the pressur- ized fluid on the thin membrane pore size. 2. Coupling-- transport mechanics predict a small coupling effect between the flow of pure water and inor- ganic ions. 3. Hydrolytic degradation of cellulose acetate to cellulose. 8 The first and last factors have already been dis- cussed above in connection with decrease of water flux through membranes. Current knowledge has not determined the importance of the second factor in salt rejection. Further research is needed to identify the magnitude of the coupling effect and how to overcome it. Membrane Transport Mechanisms The tranSport of both water and salt across a mem- brane has not been adequately described in a physical theory so that all scientists involved could reach agree— ment. Of the many theories proposed, two stand out as having satisfied most of the experimental evidence to date. These are the solution diffusion model and the pore flow model. The solution diffusion model pictures a homogeneous gel phase in which water and salts are dissolved and trans- ported by means of diffusion as described by Fick's Law(22). The pore flow model is being more widely accepted to- day. It is described thusly: "A typical membrane material can be viewed as a water swollen, rubber-like sheet of material with an active sur- face layer Of 0.1 to 0.3 percent Of the total thickness. The thin active layer contains water molecules bound to the polymer structure by hydrogen bond. Dissolved salts and organic materials are rejected at the active layer because bound water is no longer available to dissolve solute. The pure water that passes through the bound water portions of the membrane layer is transported by successive forming and breaking of hydrogen bonds between water molecules and the "active sites." Under pressure, a water molecule ap- proaches a membrane site where a molecule is already bound. The approaching water molecule forms a new bond as the pre- vious molecule is freed by that site. By site transfer, bound water "diffuses" into the body of the membrane. In the swollen, sublayer body, capillary action then moves the water molecules through relatively huge pores to a sink at atmOSpheric pressure. The dense surface layer appears to limit the flux and is responsible for the greatest pres- sure drOp across the membrane."(l4) Using this model, Kesting(23) concludes that in an asymetric cellulose, 1.5 acetate membrane, two water mole— 9 cules bound to one "active" group(an -OH group) can be regarded as primary bound water. The next 3-9 moles of water absorbed by the membrane per "active" group is sec- ondary bound water, and all water absorbed in excess of these values is basically "free" water. Research in the transport mechanism of reverse os- mosis is being encouraged by both private and government- al sources. Knowledge in this field is important to de- sign Of membranes with higher fluxes and lifetimes. Membrane Configuration There are four different basic membrane configura- tions currently being evaluated for use with reverse os— mosis units. They are: 1. Plate and frame(PF) 2. Tubular(T) 3. Spiral wound(SW) 4. Hollow fine fiber(HFF) The plate and frame configuration was the first type explored in early reverse osmosis development by Aerojet General Corporation. Current studies show it has advant- ages Of simplicity, ruggedness, and on-site replacement. But its disadvantages of high equipment cost, difficult brine flow patterns, and high labor membrane replacement costs dim its future development into municipal use. The concept is shown in Figure 3(14). The tubular configuration is similar to a typical shell and tube heat exchanger, as shown in Figure 4(14). Cellulose acetate is cast on the inner supporting wall. Its Obvious advantages are its well defined flow passages, small filtration requirements, and ease of cleaning. But it has a low packing density, with a large number of tubes required per unit surface area, and is moderately high in initial cost. Research is aimed at lowering costs by de— velOping in-situ casting Of membranes. 10 Membrane Support Plates if '11 CD (‘D Q. E I C: ’ ’l 1 ‘ E V 0 '\\Z\\\\ ITIIIIII E 111111111 'IIfIIIIIII ifmxxxxxxn‘ \ 3—; I . {3 3'"; .r-t7' ‘ a: j '3: I“ 11?”. q- --: .mr ~u-g... ‘ :(Xl ("III-4 I : r .3 § ,5 9 it #8 we ~ ' z : ~ 5 \ i l i s E i § ,3 \ ‘ I < . v \ g s ; § ,m, ‘ i § § l ‘ a .’ ‘ \ _ i 3 a s s s 5 ,_ I, e. l \\\ \\‘. * $7 \ V T\H I Reject Product ShaiItI Pressure Plate Figure 3. Plate and Frame Module Configuration Porous Support Product Water See Detail A Tube with Replaceable Osmotic Membrane * “L A A A 4 A A A ‘ oq: O 00 DO 0 0°00 00 0‘3 Do 83 3° 80° 00° °o°o° Brackish oo' o o o 0000:0093 O O o O O O O Water 0 ° 0° on. o o O o 000 00000 00 00000 00 O n0 Ooonron 0 no 9 U V U U ‘8 I3 5 i 6 5' 5 Fiberglass Product Water Reinforced Epoxy Tube Cellulosic Liner Osmotic Membrane Tubular Module Detail A Assembly Figure 4. Tubular Module Configuration 11 Development of the spiral wound design has been the emphasis of Gulf General Atomic Company. High membranes packing density is possible, and it is adaptable to sim- ple field replacement. However, the product flow path is long, and much feed water filtration is required to pre- vent plugging. Figure 5 shows a simple diagram of the module(14). Brine-Side Separator Screen Product-Water Flow \ "\. (after passage through Product Water ‘ 1 membrane) ”,2 \ ‘ s ‘ ex" Product-Water-Side Backing \ //' Material with Membrane on Each Side, Glued Around ‘ Edges and to Center Tube V Membrane Product-WaN Side Backing *7 Material ‘ Membrane A 7 Brine-Side Spacer /C Figure 5. Spiral‘Wound.Module Configuration The hollow fine fiber concept has been developed by the Dow, Dupont, and Monsanto Companies. Figure 6 shows a typical module as manufactured by Dow Chemical. The configuration is similar to the tubular design with a large number of hollow fibers of very small diameter rest- ing in a metal shell. The fibers are composed of un— supported membrano material, either cellulose acetate or nylon. ‘9 Vi ll Epoxy ‘ ll? 'l"‘:‘; VF . 1'r‘r:l;n‘,. . . . ' 4'» ‘7 a ' - ' o 4...- O) '33 ' :3 [ rt - ‘l O l' “WNW" N 0 - .- ' «a H 1 11 1 4-. 1:2.“1. 1. g £3;lilifi‘él-N E 1 'tif:s‘;;;%it..;.;31 3;“ 3 3491."? 7+“ 5 i ‘ 11::::;“{: 1- 'A m i.‘ 11 E‘ l. :1“! . ”1, 1' O. "-llllfll’l-l {N ‘lIIIII Ii' . o ‘fllthf' .. 'll' 1 ' O '.‘\:. .n."uv.- ' o Hmtlll - 1:" I f1 I CF33 Cr—33 C2223? (CZ: -B fine—+- c=n cc::=a cc:::u LA’r T vv~7 )1 WIIQF l ‘4 . v l'l a o I Z .‘r. 1 4*7 l. 2’ i"- - .1 any . 1’ A A, - l .. . . 11 ": ' 1;}! a. > ._ 1 l" - l I.' ,0 ‘I 1 . . .1. . F. 1 .11 '. '1'. 1' I IIi -' I} 1". "l" “0‘ '1 °. ”52“,“; . i ‘ *1 Illllll 1' fi§; - $57 t==33 {Cf-=3 ¢=3 Permeate Outlet 0' /2 1 C? C ===3 C :33 1' . '-/./.’-. '._i_L_°_.' % rr 42 I I A 'l' l.’ A A, . .‘- fivvg av V - . “tlfu “ . ' a . ‘ I e A, c; A I’ ‘1 n 4“ - Q ' ~ . -. _ . 0 .l . O O o ‘ '4: AAA; LA..A‘A‘A‘A“A.ALJ A. III F \\ Xx Elam: (9. Dow Hollow F'me F35” R0 M°dule l3 Water flux for hollow fibers is expressed as a modi- fied form of Equation 1(5): J1 = KAQAPE-Awl - (4) where, K = permeability coefficient of the membrane for water A = membrane are ea(ft2) t - membrane thickness(in.). This equation shows that the flux is directly pro- portional to K and A at a given temperature and inversely proportional to t. A major objective of water purifica- tion with membranes is to pack as much membrane area in as small a volume as possible. Hollow fibers allow a high packing density because no support material is needed for the membranes. With nylon fibers, as designed by Dupont, there is little membrane compaction. Lastly, hollow fiber reverse osmosis units can utilize polymers whose K would be far too low for use in the supported film type of modules. This will increase the probability of finding materials with high service life expectation. Disadvantages of the EFF modules include need of con- siderable pretreatment of feedvater, factory replacement of modules, and lack of high efficiency membrane technol— ogy to date. Reverse Osmosis System Parameters A parameter of major importance in the reverse osmo— sis process is the recovery ratio. This is simply defined as the ratio of product water to feedwater flow rate. This parameter affects the system power requirements, plant costs, and chemical treatment cost. The following mass flow equations hold true for any desalting process: Mf 2-: Mb + Mp (5) ‘Mfo = Mbe + Mpcp (6) 14 where, M = Flow rate 0 = salinity sub f = denotes feedwater sub b = denotes reject brine sub p = denotes product water. The recovery ratio is defined as: RR = Mp/‘Mf ‘ (7) Another quantity frequently important in reverse osmosis studies is the brine concentration ratio. This is defined by: CR = cb/cf (8) It can be shown that the relationship between the recovery ratio and the concentration ratio is approximately: RR 2-! 1 - l/CR = 1 - cf/cb (9) From.Equation 9, it can be seen that a high recovery ratio corresponds to a highly concentrated reject brine. Operating at high recovery ratio is advantageous for several reasons: 1. Reduced power requirements. 2. Reduced brine diaposal requirements. 3. Reduced chemical treatment costs. The factors limiting the maximum allowable recovery are scale formation and increased salt transport. It is nec- essary for a system to be optimized for maximum recovery with as little salt passage and scale formation as poss— ible. Two system hydraulics parameters of considerable im— portance in a R0 module are concentration polarization and system fluid pressure losses. Concentration polarization is said to occur when the salt concentration at the membrane is greater than that in the bulk feedwater stream(l4). The adverse effects of this phenomenon are an increase in product water salinity, 15 an increase in membrane scaling, and an increase in the osmotic pressure to be overcome. The most effective solution to this problem in a specific module configuration is the feedwater flow vel- ocity. Turbulent flow greatly decreases the buildup of salinity at the membrane interface over laminar flow(l)l261. However, high water fluxes increase polarization effects. In practice, computer programs are used to evaluate local values of the polarization in particular modules. An op- timum design must be found to include both parameters which decrease the effect, and those which increase it. The concentration polarization factor can be defined as the ratio of the salt concentration measured at the wall of the membrane to the bulk salt concentration in the brine stream(Cw / Cb)' , For both tubular and hollow fine fiber modules, po- larization can be calculated by using the following equa- tions (12) (14): ' CPF = ex (10) ‘\ + (l-*\)ex x = (0.0044J1Ngé67/f3) (11) where, CPF = concentration polarization factor f a friction factor v = average feedwater stream velocity (ft/sec) Nsc = Schmidt number ‘\ = salt rejection factor J1 = product water flux (GPD/ftz) The Schmidt number is a dimensionless quantity defined as “VDS, the kinematic viscosity of the stream divided by the salt diffusivity through the liquid. The salt rejecte ion factoritis simply the fraction of salt rejected by the membrane. Computer programs have been designed to imple- ment this equation for particular systems(12). 16 System pressure losses fall into two catagories, brine side pressure drOpQAPb), and product water side pressure dropGAPp). Therefore, the effective system driving pressure is defined by the following relationship: Peff = (P -APfL -APP) - A‘ll' (12) where, P = high pressure pump outlet pressure APTi = pressure loss associated with.concentration of feedwater stream APp a: product side pressure drop 41“ = osmotic pressure gradient. The brine side pressure drop is due to friction of the brine in the brine channel. Pressure losses are typically 5—15% of the total system pressure(l4). EXPEREZENTAL EQU Immm The hollow fine fiber reverse osmosis module used in this study was manufactured by Dow Chemical Company. The auxiliary system was designed and made by Polymetrics of San Carlos, California. The supplemental equipment was either fabricated by the MSU Division of Engineering Re- search shop or purchased outside of the University. Fig- ure 7 shows a basic diagram of the system. It must be noted that only one EFF module was used for experimentation instead of the two shown. Figures 8 and 9 show the front and rear of the equipment respectively. Synthetic brine was stored in 55 gallon polyethylene drums. It was pumped into the system through a pressure release valve which bypassed excess water back to the feed tanks. In order to minimize clogging of the fibers, a 5 mi- cron cartridge filter was used to remove suspended matter. An in-line mixing tank of 7 gallon capacity follows the filter. It is here that hydrochloric acid and radioactive tracers are thoroughly distributed in the feedwater. Feedwater was forced through the R0 module by a Gould Model 3933 multistage, centrifugal pump of 15 HP. It is water lubricated and has a capacity of 2-6 GPM, but it can be operated at lower flows by bypassing liquid back to pump suction. The maximum working pressure is 800 psi; and the maximum working temperature is l60°F. Desalination was accomplished with a single Dow CTA-A Hollow Fine Fiber R0 Module. Its specifications are listed in Table 1. Delivery of chemical treatment to the feed stream was made with a manual controlled feed pump manufactured by Precision Control Products Corporation of Waltham, Mass. It has a maximum capacity of 3.4 GPD with a dial setting from 30-100 to indicate percentage of maximum feed rate. 17 18 Empmhm mHmoEmo omho>om mo Swampsm 30am .b ohfimflh AH . "v n Rupee hpfl>flppdpnoo Emohpm oaohoom Emmhpm opsoEHom hopoEdpom O HSUOE Cm . o>Hs> M§HPMH5mom QESQ ohdmmohm SMHm mQOHm oHSQmHmQEoB Enehpm opmGS HopoE mm Hfio>Homoh Uses QESQ mafihepoa Addfimz aces menses omnmw ohdmmohm posses aw Esohpm hopsBUooh .a .0 on oE .H .a .n .H .n .m .o .6 .o .D .d Figure 8. 19 \\\‘\ \\\\\ \\\\\\ \\\\\\\\‘ \\\ \‘\\\\‘u‘\ \\\\\\‘\\\ \\ \\\\\\\ \\ \\\\\\\ \\ I l I t t Front View of Dow Reverse Osmosis Unit 20 ‘Hqugu; I ii; ll‘ .- ‘, r... .1. l 4,1 Figure 9. Rear View of Dow Reverse Osmosis Unit 21 Table 1: Dow Hollow Fiber Module Specifications Shell dimensions 6.6"O.D. x 61" long Shell material Epoxy Clad sch. 40 steel Channel covers Alumi.u. End closure Victaulic couplings taximum Operating pressure 60 psig 0 Temperature range 10 to 30 C pH range 4.0 to 8.5 Water Recovery range 10 to 90% Chlorine tolerance 1 ppm continuous Flow mode of single cartridge Radially inward Active area of single cartridge 1187 sq. ft. The system contains adequate instrumentation with a pH meter and temperature probe in the feed stream; pres- sure gauges after the filter, high pressure pump, and R0 modules; conductivity probe in the product water stream; regulating valves in concentrate and recycle streams; and rotameters for all outlet streams. . The pH meter is an Analytical Big Scale meter, Model 707, produced by Analytical Measurements, Inc. It allows for calibration with a buffer solution before use in ex— perimentation. Though not used in the present study, the Balsbaugh Laboratory conductivity meter is available for readout on several scales in the permeate stream. Start-up of the system is accomplished with either manual or automatic "on" buttons. It is better to Oper- ate on automatic so the system can turn itself off when inlet pressure drops below the minimum required for pump operation (about 10 psig). EXPERIMENTAL HETHODS Prior to experimentation, calibration of the pH meter and rotameters was necessary. The pH meter probe reser- voir was filled with KCl solution before connecting the probe to the meter. When that was done, the probe was rinsed in distilled water and placed in some pH=7 buffer. The Asymmetry knob was rotated until the pointer read pH=7. The rotameters were calibrated by timing the flow of a quantity of water into a graduated flask. It was useful to perform this Operation for every run because of the many bubbles of gas that appeared in later experiments. The first half of experimentation was made with tap water as feed. Cold water was fed directly from an indoor spigot to the prefilter at about 50-60 psig. Later, when synthetic brine was mixed and used, the feedwater was pumped into the prefilter at about 20 psig. Before start- up, an acid solution was mixed for pH control and poured into the acid reservoir(a 1000 ml graduated cylinder held by a ringstand clamp). The procedure for start-up and experimentation was as follows: 1. The brine, or concentrate, valve was opened fully. 2. Check was made that an adequate supply of acid solution was available, and that the pump was primed and leak—free. 3. The main tap water valve was opened to the system. 4. Tap water was allowed to flush the system to re- move air. This is completed when no bubbles appear in the brine rotameter. 5. The brine flow was adjusted to about 2 GEM. 6. If prior check had been made to see if motor rota- tion is correct, the high pressure pump was then turned on with the "auto" button. 7. Re-adjustment was made to the concentrate flow for 22 23 the desired rate; recirculation rate was adjusted if used. 8. The acid pump was adjusted until the proper pH of 5.5 to 6.0 was achieved. This sometimes required 30 to 40 minutes. 9. If the system was at equilibrium, flow rates were constant, and the pH was steady, the radionuclide was added to the acid supply vessel. Mixing was accomplished with a magnetic stirring rod. 10. Samples of feedwater, product, concentrate, and recycle(if necessary) streams were taken in small vials. Subsequent samples were taken at 30 minute intervals for perhaps 1% to 3 hours, depending upon how long feedwater solution or the acid solution lasted. 11. Periodic checks of the pH and temperature were made during the experimentation. 12. Flow rates were measured with a graduated flask at least twice during an experiment, and an average value taken. ' 13. When all samples had been taken, the injection of nuclide was stepped, and an uncontaminated acid solution was pumped into the feedwater. 14. The system was flushed for several hours after experimentation, usually at higher flow rates than were used during the run. 15. All power to the system was turned off, and the high pressure pump shut down. 16. Feedwater flow to the system was stopped. It was very easy to lose the priming on the acid con— trol pump. The best way to avoid this was to keep the suc— tion end in solution at all times when not in use, to take it out of solution between pulses, and to shake the bubb- les out of the suction after replacing in solution. . Synthetic brine was mixed in the polyethylene drums according to the proportions listed in Appendix B. The best method of mixing was to pour the chemicals into about 24 20 gallons of hot tap water and 200 m1 of hydrochloric acid. After stirring a few minutes with a wooden paddle, cold tap water was added to the 50 gallon mark. It was desirable to have the room fully ventilated so that acid fumes could be diapersed quickly. Safety Precautions The use of radioactive tracers necessitated some pre- cautions for safety to prevent gross contamination of peo- ple, equipment, and waste streams. A pocket film.badge was worn at all times when in both the Reactor Lab and the Chemical Engineering Lab. The equipment was always slightly contaminated after the first use of radioisotOpes. Absorbent paper was taped to parts of the R0 unit as well as to the floor near the out- let streams. This was discarded frequently to reduce con- tamination from leakages or spills on the equipment. Plastic gloves were used when preparing the radioact- ive sample in the Reactor Lab. But if one was careful when taking samples during experimentation, no gloves were needed because hands and sample vials would remain uncon- taminated. Contamination wipes were taken during each experiment to determine if the equipment was becoming too active for safety. All of these samples were low in activity except when.a drop or two of acid solution would leak from the acid pump. Appendix C shows the activity of each contam- ination sample. w , All sample vials and glassware were thoroughly cleaned with soap and hot water after each experiment. Contamin- ated liquids were disposed of in the proper radioactive container. Most of the radioisotOpe samples used had theoretical activities between 20 and 45 microCuries before mixing with the feedstream of the R0 unit. Samples of S35and Ca45 were usually 20}iCi in strength, while more of Na24 was used in 25 experimentation because of its much shorter half-life. The entire amount of the radioactivity was never used. The acid reservoir was a 1000 m1 graduated cylinder into which the suction end of the acid pump was dropped. It required a minimum of about 100-150 ml of solution to function. Therefore, out of a 1000 m1 of acid solution, perhaps 10% of that solution(and radioisotOpe) was not used. The radioactive sample used in the system which was not retained in the system.or taken out with samples, was released to the sewer. This did not constitute any health problems because of the amount of water released with it. During experimentation with tap water, the system used about 100 gallons of water per hour, diluting about 1/3 of the original amount of activity(6 to 15;LCi). This is based on operation of the system for 2 hours for experi- mentation and at least an additional hour for flushing. With the synthetic brine operations, the feed rate was reduced to about 75 gallons per hour because of the lower feedstream pressure. Dilution was slightly less be- cause an experiment was run in 1% hours. But, more time was spent flushing the system than with the tap water feed. ANALYTICAL METHODS Liquid scintillation spectrometry was used to measure the radioactivity of experimental water samples. This me- thod afforded a fairly efficient, consistent, and time saving analysis of the beta-emitting nuclides studied dur- ing this research. The system used was a Packard Tri—Carb Liquid Scin- tillation Spectrometer System, Model 3003. The details about scintillation theory and about a typical scintilla- tion system can be found in Appendix A. Preliminary ad- justments had to be made in the system before it could be used. First of all, the temperature control of the count- ing chamber was adjusted until the chamber temperature stabilized at about 5°C. Secondly, a high activity sample of the nuclide to be studied(S35 , C345 , Na24) was used to determine the maximum.gain for each nuclide in one or all channels. This was done by counting the sample for a minute at different gain settings and plotting counts per minute(cpm) vs. % gain setting. Appendix D contains the results of these plots, as well as calculations for the detector efficiency. The particular Spectrometer used is convenient for sample analysis because it has a sample changer with a ca— pacity of 200 sample vials. A sample is counted for a preset number of minutes or counts. Each successive sam- ple is automatically counted by the detector after one, or both, of the limits is reached. The sample number, time elapsed, and counts in all three channels are printed out by a digital printer on paper tape. In this study, radioactive samples of S35 and Ca45 were detected by mixing 9 m1(tl.ml) of the aqueous sample with 9 m1 of Packard Insta-gel Emulsifier in a polyethylene vial. This was placed in the sample chamber(after vigor- ous shaking) to be counted for 50 minutes, or 100,000. 26 27 counts, whichever limit was reached first. An empty vial was counted for background. It was convenient to let the samples count overnight. Therefore, samples from each ex- periment were counted at least twice. 24 were detected with the same spectrometer, but using a different tech- nique- Cerenkov radiation. This method requires no scin- tillation liquid. Eighteen milliliters of the aqueous sam- ples were poured into polyethylene vials and placed in the counting chamber. The vials held 20 ml when full, and mea- surement of 18 ml into several vials provided experience for pouring the rest of the samples. This method intro- duced some error into the activity readings of the samples. The quantity of liquid in the vials may have differed from 18 ml by t 1 ml. The same time and count settings were used, but the optimum gain settings were changed. Samples were discarded in the contaminated waste receptacles after counting. The high-energy beta emissions of Na DATA All experimentation was carried out within a pH range of 5.7 to 6.1. Within these limits, membrane deteriora— tion was kept to a minimum and the high concentration of calcium, sulfate, and chloride ions did not cause large amounts of precipitation. The performance of the system was studied by measuring the relative activities of feed, permeate, and concentrate streams. This data led to re- jection rates over a period of time for particular ions such as 8032, Ca++, and Na+. Stream flow rates were also measured. Raw data is presented in Appendix C. Temperature and pressure data were gathered for each experiment. The system pressure changed very little dur- ing experimentation(staying near 700 psig for tap water, and about 670 psig for the synthetic brine). By increas- ing the concentrate or recycle flow rates, the concentrate pressure was lowered to as low as 510 psig. The feedwater temperature(which is near to the Operating temperature) was recorded so that fluxes could be corrected to 25°C (77°F). This was done by using a chart provided by Dow Chemical Company listing correcting ratios to 25°C for temperatures from 100 to 30°C (Figure 25, Appendix D). Appendix D includes methods of calculating rejection, recovery, water flux, and standard deviations. Experimentation was divided into two segments. The first part utilized tap water as a feed stream. It was in- tended to be a period of learning preper techniques and of determining some of the variables of the system. The more important portion of the experimentation was use of a syn- thetic brine of roughly 8700 ppm.as a feed stream. For this stream, rejection was measured for so;2 and Na+, wat- er flux was calculated, and overall behavior of the R0 unit was observed over a period of time. The experimental data derived from the experiments is 28 influh ‘:—— drlfl‘u -. ++FHHIU quake dlfli,-. icy...“ tllllvi unli-i- Ii.i-FH c-1lyiflh C li.‘lll.‘ltl‘fll.'llhl.rufi I!“ HI.I.IN.I.. I. rl. 29 .An use b modems oomvsopmsm one :a hpe>auoe Hessflmmh 03p ofiflshmpsv 0p mmmflnmdflh Eopmhm who? m was .b.w.: mpdoafihmmxm nosoz n.Hm n.mm e.rw w.ne Nw.a na.a 03.0 mm w.n mmw mom ea m.mm n.mw o.me Hw.a NH.H 0:.0 rm m.n mam mos ea soda oaohooa 2mm ow.o n.nm n.ww o.ee me.a :H.H 0:.0 an m.m con can ma n.rm N.am 0.:m o.ss Hw.a sa.a 0:.0 mm w.n cos can as e.:m o.ww w.nm N.sn Nw.a na.a oa.o mm w.n 005 can ma zodk odohoeh oz COH ++mo H.mm n.em w.mm m.am we.a mo.a oo.a mm m.n was now as n.wm w.nm N.wm N.Hm mm.a mo.a oo.a an o.m mow mom as Rom n hhm>oomm m.mm m.mm m.em w.ma ow.a ma.a or.o so m.n mam non oa H.5m H.mm r.mm m.mm m.mm an.a oa.a 0:.0 an m.n 005 can m m.nn ~.ow o.mm N.:s me.a ma.a 0:.0 ow m.m con mos n m.aw e.am a.ow m.ma me.a oa.a os.o mm m.m mam mos m o.ee m.me o.me s.ma em.a mo.a os.o mm w.n cos mos N m.n w.ae om.a ma.a ms.o mm m.m mam mos a RmBIOh n kho>ooom soa Nmom om» oma om ow :ee Tea: 25 3; mam Ema Amamdv Aewev ans 00mm spam 3oam Ammovmpmm Amov .mmmhm .mmoum .cz Compoehom COH R Ibooom MSHh mvmoshom 30Hm .ofioo .QEmB ma .fimocoo anmhm .mxm mGHmD COH .7 +60 Udd GOH .Edohpm book we Housz awe mom how sonfispno span Hepfioafihomwm «N manta 3O .A.GHE om\oooo: asap mmoHVmOHmEsm on» as kpfl>flpom0flflmh SMSonm pom mm3 cheap #59 .UopQEoppw ms? wH pfiofiflhomxm «mpoz m.wm m.mm w.om m.ma we.a NH.H oa.o aw o.o mam mos am w.nm n.na e.nm m.ms ee.a NH.H 0:.0 mm m.n mam mos mm roam machete mas ow.o e.mm :.nm N.mw mane mw.a nH.H 0:.0 mm n.n mam mom mm ~.nm m.~m N.ww w.me ms.a NH.H 0:.0 ow m.m mmw mos am m.mm m.nm H.nm m.me me.a NH.H 0:.0 ow m.m mam mos om N.mm n.ms sn.a oa.H 0:.0 mm o.w mam mom ma BOHH oaohoeh 02 0mg .oma so am A.eaeleeae has med grade Amamev Amaeev awe comm deem soda Ammovoumm Aaov .emoam .mmoum .oz Gospomnem GOH R I>oosm Nddm cpsofihom 30Hh.0:oo .QEeH mm .flsodoo Eovuhm .mxm mzflmb GOH + .Bsohum doom up Hope: use $2 How sodfispno span Harneafihomxm uh Odnse 31 30.0 00.0 30.0 00.N an om.o mm.0 hm.0 nh.d mm Hm.0 00.0 00.0 00.0 mm 00.0 00.0 00.H 00.0 Hm 50.H Nm.0 0:.0 00.0 on NH.H mo.a Mb.0 00.0 mm 00.H 00.H 05.0 00.0 mm mw.0 00.0 00.0 mm mn.o mm.o 00.0 mm Hm.a Hm.0 00.0 mm om. 00 on . A.GHEvoEHB Azmwv .oz oavmm <0 Guam OHOhomm .mxm 0.00 H.Hm :.mw N.wh H0.H 00.H 0N.0 mm 0.0 mn0 0B0 an n.om n.ww N.om m.me No.a mm.o mm.o om 0.0 ore one no 30Hm mauhooh Sufi? N.mw N.nm n.nm b.Hw 00.0 nm.0 HN.0 mm 0.0 M00 0&0 mm n.om m.ww m.ww e.w0 NH.H 0m.o e:.o ow m.m 000 one an H.00 H.Nw m.mm m.Nb N0.H mm.0 wm.0 00 0.0 .000 050 on m.mw N.mw 0.50 N.Nb 00.H m0.H 03.0 om 0.0 000 050 mm o.nm m.mm e.aw m.m0 mo.a em.o m:.o mm «.0 mme mum mm n.mm n.ww o.r0 mh.o mn.o me.o mm m.m moo new em m.hw m.mm H.00 MH.H Nm.0 H:.0 00 0.0 000 mh0 0N m.mw m.Hm H.mw ma.H 0H.H ma.0 Om 0.0 n00 000 mm 3oah oaohowh oz om 00 0011 19% 7:333 as mam :33 A330 Amanda has 000m ovsm 30Hm Azmwvoesm Amov .mmth .mmon .oz Gowuoonom SOH R I>ooom Nada sumoEHom 30Hh.ofloo .0608 m0 .Goodoo Eopmhm .mxm .Esohpm pooh ms ocdhm oasonufihm wnflmb SCH +62 Rom vsnfimvno span stflmafihomxm u: OHQMB 32 00.0 oo.0 00.0 a: HH.H 0.0H 00.0 0: 0H.H 00.0 00.0 00 00.0 H0.H 00.0 00 00.0 00.a 00.0 00 00.a 00.H 00.0 o0 00.0 00.0 40.0 00.0 00 00 0o 00 A.:asvoaae A2000 .oz capmm <0 oumm odohoom .mxm 0.00 0.o0 H.00 00.a ~0.a 00.0 00 0.0 0H0 0oo a: 0.o0 0.00 o.00 00.H 00.a 00.0 00 0.0 0H0 00o 0: roam oaohoop noes 0.00 0.00 0.00 o0.0 00.0 00.0 00 0.0 0oo 00o 00 0.00 0.00 H.o0 00.0 a0.0 00.0 00 0.0 0oo 00o 00 H.00 H.H0 0.on 00.0 00.0 00.0 00 H.o 0oo 00o 00 0.00 0.00 0.00 00.a 00.H 00.0 00 0.o 0oo 00o o0 0.00 0.00 o.a0 0.00 00.a 00.0 00.0 00 0.o 0oo 00o 00 30am odohooh oz 00 0o :00 A.:aevosae A00 mmm A2000 Amemav Amamav as oomm opom toga Azaovooom Ame .omosm .mmoam .oz Gospomnom GOH R I>ooom 2500 ondafihom 30am.ocoo .QEoB mm .aoofloo Eopmhw .me Edohum pooh mm ocflhm oasonsdhm muses soH «:00 tom oooampso been Hepcosesoaxm ”0 oases 33 tabulated in Tables 2—5. The tabulations include experi- mental conditions such as temperature, pH, and flow rates, as well as the system parameters such as ion rejection, water flux, and water recovery. Table 2 contains data ob- tained from the tap water experimentation for the so;2 and Ca++ ions. In Table 3 is the same type of data for Na+ ion in tap water. Tables 4 and 5 list experimental data for the so;2 and Na++ ions in the synthetic brine solution. Certain experimental data was not tabulated either in the raw data of Appendix C or in the data tables of this section. It was noted in early experiments with sulfate ion that some radioactivity remained in the module after the radioisotope was no longer being pumped into the system. In Experiments 4,6,7, and 8, attempts were made to deter- mine how much activity resided in the module. Experiments 4 and 6 produced sample activities far above the concentrations of radioisotopes used up to that time. It was concluded that either the samples had been contaminated while being prepared for counting, or the scintillation spectrometer was working improperly. Exper- iment 7 was a flushing of the mixing tank by forcing tap water through the tank and draining it before it reached the high pressure pump. The activities are shown in Table 6. . Table 6: Activities of 9 m1 Samples of Mixing Tank Flush. Time(min) Net Counts/50 min. (corrected for background) 00 1754 60 1918 . 90 1739 150 1693 210 3223 270 1668 Experiment 8 was a flushing of the R0 system to see when the residual activity would reach a steady value. The activities, corrected for background, are listed in Table 7. 34 Table 7: Data Obtained from System Flushing to Determine Residual Activity. Permeate Rate = 1.10 GPM Temperature = 59°F Concen. " = 0.87 " pH = 6.0 Recycle " = 0.00 " System.Pressure= 705 psig Time Feed Permeate Concen. (min.) Activity Activity Activity (c/50min) (c/50min) (c/SOmin) 15 3801 1423 8764 60 2571 1474 5414 120 2795 1183 1170 180 1174 1093 1190 240 1229 1226 1215 The activities in all three streams seemed to reach a steady value after 4 hours. Because of this substantial residue in the module for so;2 ion, the experimental data had to be corrected by subtraction of stream activities taken at steady state after flushing. This is essentially a background correction. 'The problem of a residue for Na+ was small due to its short half-life. A couple of hours flushing after an experiment seemed to remove almost all radioactive Na+ ions. RESULTS The raw data contained in Appendix C was used to cal- culate the percentage of salt rejection by the R0 module. This was done by using the equation: % Ion Rejection = 100(Feed Activity-Permeate Actt) (13) Feed Activity Rejection was determined at different time intervals in hopes of showing an approach to steady—state for each par- ticular ion. Consistency in the rejection results is fairly easy to see with the use of tap water. Inconsistent results can be discarded on the basis of experimental conditions and the raw activities. However, it was not as easy to find consistent results with the synthetic brine. Therefore, a parameter was designed to test the ac- curacy of results. This is arbitrarily called the CA Ratio. Its rigorous definition and a sample calculation are given in Appendix D. In brief, this ratio compares two products: the product of the concentrate flow rate and its activity as determined by material balance(Equation 6) to the same product as determined by using the experimental flow rate and activity for the concentrate stream. Ideal- ly, this ratio should be equal to 1.0. A value less than one indicates that more radioactivity is leaving the sys- tem than entering(theoretically impossible). A better ex- planation is inaccurate feed activity data from imprOper ' mixing in the R0 system or errors in scintillation coun- ting. A CA Ratio greater than one indicates some of the nuclide is staying in the system, perhaps as precipitates, or that the scintillation counting of the concentrate stream is in error. The standard deviation of the ratio is about 0.05(based on typical activities of 10,000 c/ 50 min). Water recovery was an important system parameter to calculate. The percentage of water recovered by the R0 36 system is simply the fraction given by: % Recovery = 100(Permeate Rate) (19) Permeate Rate + Concen. Rate The denominator is actually the feed rate. For the sys- tem.under study with one HFF module, the nominal design feed rate is 1.62 GPM(2332 GPD)(21). Water flux is the permeate rate in GPD divided by the active membrane area Of the R0 module. For the Dow HFF module, this area is equal to 1187 ft2(21). Example calculations of both water recovery and water flux are included in Appendix D. Results With Tap Water The first half of experimentation utilized tap water as a feed stream. This was convenient because indoor spi- gots in the Chemical Engineering Lab delivered water to the system at a high inlet pressure Of 50-60 psig. System Operation gives the best results if the inlet stream is pressurized to at least 50 psig. Analyses of the tap wa- ter indicated it contained 350-370 ppm of inorganic ma- terials. Three different ions were studied in the tap water, sozz, Ca++, and Na+. The ion rejection approach to steady state was studied for each ion. Differences in recovery were investigated with the sulfate ion, while the use of the recycle stream was studied with calcium.and sodium. Figures 10 and 11 show the approach Of so;2 ion re- jection to steady state for 70% and 50% water recovery re- spectively. These plots are Obtained with data points from.Experiments 9-12 primarily. Previous experimentation did not appear to give consistent results due to several reasons. These were inexperience with sample preparation for scintillation counting, improper scintillation coun- ting conditions(% gain, temp., etc.), and lack Of reaching steady radioisotope injection into the R0 system. The point at 150 minutes in Figure 11 is inaccurate_and should not 37 E00900 000m us hede 008 mnfimb kho>ooom Hmum3 R05 #0 SCH Nwom HOM mEHB .m> :OHpoohmm 00H R mo pOHm .OH GHSMHm A.GHEVoEHE OmH OmH ONH Ga 90 fin 0 . , . , . w 00 .00 as a M L. 4.3 H e H .OOH uotqoefea uoI % 38 3mm;- 800900 0000 00 #0003 008 wGHmD 0h0>oo0m H0003 Ron #0 QOH mmow H00 0EH8 .m> GOHpo0h0m GOH R MO pOHm .HH 0H5®Hm A.SHEV0EH8 000 000 00 0o 00 S 1 d J‘ 0w stOOH notisereu UOI % 39 give the appearance Of an increase in the curve. The CA Ratio of this point is about 1.25. The experimental data indicates that the permeate activity is low. This may be due to error in sample preparation. Figure 12 shows the results Of the effect of water recovery on % ion rejection for so;2 ion. As the amount of water recovered as permeate increases, the amount of ions rejected by the membrane decreases. Figures 13 and 14 show the change in % ion rejection with time for CaJ’+ ion at 60% recycle and no recycle re- spectively. Two data points were found to be inconsistent and were not plotted in Figure 14. These were the 88% re- jection Of Experiment 13 and the 68% rejection of Experi— ment 15, as listed in Table 2. They were not included on the plot because their CA ratios were very far from 1.00. The results Of experimentation with Na+ ion in tap water are presented in graphical form in Figures 15 and 16. The first Of these plots shows the approach of ion rejec- tion to steady state with a 60% recycle stream used. All of the data points from Table 3 are plotted because of their close agreement. Figure 16 contains a graph Of ion rejection vs. time also. But the experiments involved used no recycle stream. Experiments 21 and 22 contain the most consistent data. Experiment 19 was rejected because the low sample activi- ties contain large statistical errors. Data from.Experi- ment 20 seems good after 60 minutes having passed. The position Of the curve at 60 minutes is debatable because the data lists 3 points within a spread Of 8%. ‘Rasults With Synthetic Brine The experimental results obtained from the synthetic brine are not as easily analyzed as those from tap water. It was more difficult to find consistent data with % re— jection vs. time. There were definite trends, however, that deserve the discussion given in the next section. E00Hpm U00m m0 H0003 Q08 MHHMD .GHE ONH H0pw< 00H :00 HoM 0H0>oo0m H0p03 R .m> 00000000m HOH R MO scam .NH 0H5®Hm N' 3H0>000m H0903 R 00 00 0.0 0o 0.0 0.: 00 n i 0 i . , ea :00 O A. O :00 L: up 0 fill {00 H Adm uotqoefou UOI % 41 oaosoom moo mess :oH owfl OmH + E00Hum 000m 00 H0903 008 00009 +00 H00 0808 .m> coapo0h0m GOH R MO 900m .MH 0H3®Hm A.fiHEV0EH8 .7 L ‘1 4? I one 00 0o 00 rLo—4 0w my .00 l O U no 0 nL. 8 0 4 TL. 0 400 u :mm A00H 42 oaosoom oz mess :oH 000) 00% + E00Hpm 000h m0 H0p03 Q08 wfiHmD +00 H00 0808 .0> HOHpooh0m GOH R 00 poam .JH 0H3®Hm A.fiHEVOEH8 90H 00 0o 00 a d 5-4—4 r—a—a cm 00 l 1 Am +03 uotqaafag uoI % 43 E00Hpm 0000 m0 H0003 008 mSHmD 0H0300m R00 £003 SCH +02 HCH 0508 .m> Goapoon0m SCH R Ho yCHm .mH 0H5®Hm A.CHEV0EH8 00a 00a 00H 004 00 00 1 q 1 d 4 Ffia~—i 8—4—4 r—e—a H—q-fl 0w om toos. uotqoefeg uoI % 44 E00Hum 000m 00 H0003 Q08 MdeD 0H0000m CZ QHH3 +02 H00 0EH8 .m> GCHHC0n0m SCH R 00 HCHm .0H 0H5®Hm A.HHEV0EH8 Dams 0.0a 0.3.. 0 .0 0o 0.0 .0 . 00 .1me dp I O u U 0 0|. 0 O :00 m. 0 u we :0 0 300H 45 Figure 17 shows the change in Na+ rejection with time where there is a high recycle rate of concentrated stream back to the feed. The data is quite consistent with CA ratios near 1.00 in most cases. The points at 60 minutes show the greatest deviation from the curve. The rejection of Na+ is shown to change with time in Figure 18 also. In this case, there is no recycle stream. Attention was given to the CA ratio of the data points listed in Table 4 to determine the most consistent data to plot. The deviation in this case is greater than with Na+ in tap water, but the shape of the curve can still be discerned. Experimentation with Na+ was carried out at several values of water recovery. The relationship of water re- covery to rejection at a steady state value is shown in Figure 19. The lone point at 85% recovery is neglected in drawing the curve because of its high CA ratio(l.3l). The most perplexing part of the research endeavor is shown in Figure 20. The rejection of so;2 in synthetic brine over a period of time is unlike all other results. Much of the data collected for this ion was not accurate (if the CA ratio is a good basis for accuracy). Experi- ments 35—37 appear to give the best results, though cer- tainly they are not consistent enough to draw conclusions about the SO42 ion rejection in the brine. 46 Esohum comm mm odfihm OHQonpfihm MfiHmD oaohoom flow #6 HOH +mz 90M oEHH .w> fiOHpoofiom 20H R ho poam .hH oHSWHm A.2HEVoEHB oma ones owa cm 8 pm 0 a v + . t . ow ntmw do; I m vd A eom m. a a O 1 TL. 0 H :mm .53 47 em oaoeoom oz sea: onflhm oauoflpfihm 2H hho>ooom Ron «2 ROM oEfiH .m> cospoonom GOH R we scam .wa ohsmflm + A.nHEVoEHB bma on on on 90 o ow lop Av mm I m e m e w. w a 1- a m m. to u J,mm ..OOH ocflhm caponvfihm mawmb .GHZ ow hop%¢.flOH +wz MOM hho>ooom Hops? R .n> :cwpoonom QOH R Mo aOHm .mH chamflm hho>ooom hope? R 304 pm} aw. Dbl we on J J. 0m 48 OOH notioefeu uoI % 49 \. a _ or. .. - ,n J z t L" rung.) ,Wmt.3.tfl?~..hum‘. A I: Edohpw poem on osflhm owneflufihm WGHmD SCH Nmom MOM Andonfiflpdoomwcvoafie .m> GOHpOOfiom GOH R we poam .ON oHSMHh A.EHEVoEflB can om: Don Ohm owd pm 0 . m . . . . on to: H Lyon nVOW L‘ON. m e m tom looa uotqoefeu uoI % ANALYSIS OF RESULTS Tap Water The results shown in Figure 10 indicate that steady state rejection of 8012 ion is reached after about 90 min- utes. The RO module removes 95-96% of the sulfate ions in the tap water(about 14 ppm). This rejection occurs when the system is operated near 70% water recovery. If recovery is reduced to 50%, the steady state re- jection increases, as Figure 11 shows. The increase is slight, say to 96-97%, but this increase would be signifi- cant if the feed stream contained a higher concentration of ions, and if a large amount of water is treated per day. Not only does the steady state rejection increase at lower recoveries, but steady state is reached at an earlier time after start—up. Steady state occurs near 60 minutes for 50% recovery, but not until 90 minutes for higher water re- covery. 7 Figure 12 gives a better picture of the relationship between rejection and recovery than is obtained by examina— tion of Figures 10 and 11. The result in this graph is what the literature predicts(12)(14). Lower water recovery indicates that more of the feed stream is being wasted. If the permeate rate remains constant, a lower recovery means a greater feed rate into the module. This increased velocity creates more turbulence for a reduction of concen- tration polarization(and salt transport through membrane). In some applications of reverse osmosis, however, the op- eration is designed to proceed at as high a recovery as possible to reduce Operational and waste disposal costs. The slight loss in rejection can be tolerated in such cases. When waste is not recycled into the feed, the rejec- tion of Ca++ ions at steady state is about 93-94%(reduc- tion from 80 to about 75 ppm for MSU tap water). Figure 14 indicates that steady state is reached near 90 minutes after start-up. 5O 51 As the feed stream becomes more concentrated from.re- cycling part of the waste, the rejection drops. For 60% recycle, Figure 13 indicates that it drOps to 92-93% and does not reach steady state until 120-150 minutes. The in— creased salt flux is predicted by Equation (3) J2 = B(Cw - CP) The concentration of the feed stream at the wall Cw may be increased as much as 2-3 times what it was origin— fee ,ally. It is interesting to note that though Cw_may be in- ‘ creased considerably, the rejection of ions is reduced only a few percent once steady state is reached. 5 For the Na+ ion, the use of recycle has an opposite '_ effect upon ion rejection. Figure 15 shows steady state ' fig rejection of Na+ to be 95-96% with 60% recycle. Rejection ' is shown to be only 93-94%, however, in Figure 16 when no recycle is used. The explanation for this phenomenon is difficult to come by from looking into the literature for similar results. The literature does indicate that under identical operating conditions, Na+ ions are not rejected as easily as Ca++ ions. By comparing Figures 14 and 16, one can see those results(12). Hypotheses might be forwarded to explain the Na+ ion behavior, but none can be proven at this time. One is that the increase in Na+ concentration with recycle increases the amount of Na2804(which is more highly rejected than Na- Cl) (12). The same behavior would not arise with Ca++ ions because CaSO4 is not rejected as easily as Na2504. Another hypothesis is that the data taken was in error. But the apparent consistency of the results in this case has al- ready been discussed and taken to be a fact. The slight difference in values of water flux may be due mostly to error in taking flow rate measurements. The graduated flask allowed approximation of the flow rates to t0.05 GPM. Subsequent error of i0.05 was introduced into - t ‘7'.“ the values of water flux. 52 There was generally a slight decrease in water flux when water recovery was decreased. This is explained by the feed rate being above the nominal design rate of 1.62 GPM. With more water going into the module than it was de— signed to receive, less water is forced through the mem- branes. There was no appreciable change in water flux with use of the recycle stream. gynthetic Prinz t was originally desired to determine the rejection of several ions in the synthetic brine. But time permitted the investigation of only two, Na+ and SOXZ. Both were studied with and without the recycle stream. No attempt was made to look at the effect of low water recovery(50%). It is believed that the economics of municipal use of re- verse osmosis will require operation at high recovery rates. Rejection of Na+ ion with high recycle rates is shown in Figure 17 to reach 90-91% at steady state. This point is reached quite early in the experimentation. However, when no recycle is used, steady state rejection is still near 90%, but it is not reached until after 90 minutes of operation. An explanation may be that the increased turb— ulence in the feed stream reduced concentration polariza- tion to such an extent that high rejection is possible soon after start—up. An ion rejection of 90% is quite good for a brackish solution of 8700 ppm concentration. But this amount of purification does not produce potable water. The product stream still contains about 870 ppm of inorganics. The Michigan Department of Public Health indicated to the aut— hor that drinking water should contain less than 300 ppm inorganics. When recycled concentrate is mixed with feed, the mixture concentration may reach 2-2% times the origin- al concentration. The product stream will contain more in- organics in it as a consequence. There is a decline in ion rejection as the water re- it.‘ lam-MC ,A- ‘iduL-‘mw—j . 0 ’ n t ' r- is}? u—‘1: r. 53 covery increases. This effect is shown over a range of about 20% recovery in Figure 19. With the water flux re- maining nearly constant, the high water recovery allows concentration polarization to become a large influence aur- onn:rejection of ions at the membrane wall. Equations 10 and 11 show that the effect of concentration polarization is increased as the average feed velocity is decreased(as in high recovery). Conclusions as to the rejection of SO; reached at this time. The data was either inconsistent or gave results that show a failure of the R0 system to func- tion properly. In Figure 20, rejection data is plotted as a function of time over several experiments. The loss in rejection ability may be due to several reasons. First, after about 10 or 12 experiments with the synthetic brine, the module may need a thorough cleaning. Deepite pH control near 6.0, the high concentration of in- organics may have caused seme precipitation of sulfates, chlorides, and carbonates. The Dow Chemical Company should be able to give specifications as to how cleaning should be done. Secondly, some of the cellulose acetate may have de- teriorated to cellulose, leaving "holes" in some of the fibers. The large concentration of inorganics in the per- meate stream supports this idea. However, when recycle is used, the rejection becomes much higher again. Lastly, the concentration polarization affect may have become so great from the precipitates going back into sol- ution that salt transport increased sharply. It was re- duced with the added turbulence of recycled water. .P‘ 2 ion cannot be .1 a “tuna-u'ix'mm‘amm . . . . o .. . l.’ {5" .‘ . i CONCLUSIONS AND RECOMMENDATIONS Basic Observations The experimentation with reverse osmosis itilized two types of feed streams, tap water with 350—370 ppm inorgan— ics and a synthetic brine with a concentration of 8700 ppm. Several parameters of the system were considered, including water flux, water recovery, and ion rejection. 0f the many ions in both feeds, sogz, Ca++, and Na+ were studied in par— ticular. In both cases of tap water and synthetic brine, a steady state ion rejection was reached after at least 90 minutes. For the tap water, it was reached sometimes sooner than that. When a recycle stream was mixed with the feed, the ion re- jection decreased slightly for all cases except Na+ ions in tap water. As the water recovery was decreased by passing more water to the sewer, the rejection increased because of the increased turbulence near the membrane walls. However, the water flux decreased slightly in that case. The differ— ence in ion rejection between the two feed stocks was only 5-6% at about the same recovery of 70-75”. System performance remained constant with the tap water, primarily because of the low ion concentration. When the brine solution was used as feed, however, the performance drapped after a period of time. This fact is evident from the so;2 rejection results in Figure 20. Water flux was noticeably less for the brine than for the tap water. This drop was caused by two factors: the smaller inlet feed pressure and the larger osmotic pressure of the brine solu- tion. Further testing will be necessary with the synthetic brine(or brackish water from the Marshall formation, if pos- sible) to draw permanent conclusions as the use of reverse osmosis in Lansing. But this study shows the possibilities of 90% or higher ion rejection for brackish water at high 54 55 water recovery. At this rejection rate, drinking water could be obtained with two RO systems in series. The first system would handle the pretreated brine, and the second would purify the initial permeate to potable levels. Operation of the Dow HFF system was quite simple once procedures were developed. The necessity of a module clean— ing procedure was realized when later brine experiments showed poor ion rejection results. Flushing after experi- mentation removed most of the soluble ions, but other tech- niques are needed for removal of insoluble precipitates in the hollow fibers. _§uggested Research There are several areas that should be investigated in the near future: 1. Using the synthetic brine, the 30;2 ion rejection should be measured again. 2. Operation should be tried at lower system pressures, say about 300-400 psig. 3. Tests should be performed on.Marshall formation water for rejection and lifetime performance of the R0 module. 4. A study should be done to optimize the Dow unit in regard to recovery vs. rejection in terms of power, pre- treatment, and waste disposal costs. Suggested Experimental Improvements The pressure relief valve on the synthetic brine de- livery pump should be removed. It was initially thought that the pump would deliver too much feed to the system. But it appears that the system would operate better if more is delivered at pressures up to 50 psig. Experimentation should be taken more often than in this study to show more accurate results. The CA ratio used to test result consistency may not be entirely accurate. A better technique for determining consistency of results should be designed. 1.7.65.1! ,nWJV -' .:. net-u. .. a. BIBLIOGRAPHY BIBLIOGRAPHY l. Bixler,ft,and Cross. R. "Final Report on Control of Con- centre tion Polarization in Reverse Osmosis Desalina- tion of Water." R. &D. Progress Report No. 469 (Dec., 1969). Office of Saline Water, Department of the Interior. 2. Bransone Jr., E. D., ed. The Current Status of Liquid scantilet on Counting. Greene a Straton, New York, 19FO. 3. Chase, G. D., and Re binowitz, J. L. "Scintillation Techniques and Nuclear Emulsions." Principles of Radiol AoomN Op c>fluwdmvaSHm cumehcm Ho poam .mN whzmfim (343/ads)xnta okfiutkoQEoB mm em as we on on mod: mm . on mm om ma be com J 1 ow.o 26m.o .bo.H +bH.H +bN.H :0m.a +0.: . H 29m.a :ow.a