Desalination . 18 (1976) 99-111
© Elsevier Scientific Publishing Company, Amsterdam - Printed in The Netherlands
HYBRID REVERSE OSMOSIS - ULTRAFILTRATION MEMBRANES S. B. SACHS, E . ZISNER aND G . HERSCOVICI
'ring, Tel-.4 ri r, (Israel)
(Received June 8, 1976)
A new group of asymmetric non-cellulosic membranes having a performance intermediate between that of a conventional RO membrane and an ultrafilter has been developed . Its main characteristics are high fluxes (4-IO m 3/m 2 day) at low pressures (6-10 at) and moderate rejections for various salts . From 200 mg/I solutions of NaNQ 3 , Na_SO, and KH Z PQ t rejections up to 40 f%, 70`,' and 90%, respectively, were obtained . The new membranes can withstand large variations in pH (I to 13) and have excellent chemical and biological stability . The membranes have been tested in a mobile SUF pilot plant operating on oxidation pond effluent . High rejections for BOD, COD, bacteria and suspended solids as well as a 20"„ reduction in salinity have been obtained . INTRODUCTION
For many years membrane processes have attempted to cope with the task of providing an answer to the ever increasing water demand and dwindling water supply in many parts of the world, so far with only partial success . RO, and to some extent elect rodialysis, have been able to deal with brackish water, some industrial effluents, and, in a limited way, with sea water . However, the enormous quantities of municipal sewage . which represents one of the potential water sources best suited for extending limited water supplies, are still awaiting a reliable and efficient process to turn them into acceptable effluents . The new rigid standards of waste-water disposal provide an additional reason to the search for such a process . The nature of municipal sewage - large volumes of water containing suspended colloidal particles, undesirable dissolved organic matter, bacteria and a higher than permissible salt content - sets special demands on the membrane and the process . In order to apply a membrane process to the treatment of municipal sewage . a membrane which exhibits high water fluxes, rejects organic matter and bacteria, as well as part of the salts, is needed . In addition, the membrane should be chemically stable towards the components of sewage and maintain its properties over extended periods of time .
S. B . SACii5 . E. !_I5%ER AND G. HLRSC <311C1
Besides the need for a special membrane, the treatment of s~ev~age also requires certain modifications in the existing membrane proces=ses . While in RO units the hi9My-effidert Salt the-ring required demands relatneJy dense membranes and high pressures . ultrafiltration is based upon open_ porous membranes which should operate at low pressures and reject high molecular vbeight molecules only . Roth processes and the membranes used in them are not adequate to deal with the barge amounts of particulate matter present in municipal snA age arid experiments with RO systems on these effluents ase not been very successful 41) . Thus . none of the existent processes is suitable for municipal snage treatment, We ha-.e deveicped a process v .hich is intermedia-e between RO and ultra.agr ultrafiliration (SUF) . filtration and have termed it tev In SUF- in order to present the dogging of the membrane by accumulation of filter c :ake . relatisely high axial feed seloc-ties past the membrane and Large flow channels compared to other processes art required . The most suitable configuration i, tubular . New- high performance hybrid ultrafil that appear to be ideally suited to the SUF process . have been deseloped at the Israel Desalination Engineering Laboraton . Because of pending patent considerations. the exact nature of the membrane material is not disclosed in this paper . We refer to them as HUF 22110 and HUF 33040 . and they have the following characteristics-(I) The- are non-cellulosic and ezceptiorwllti resistant to chemical and biological attack . They are stable in acidic and basic media_ beiwcen pH 1 and 13 . 12) Thes withstand elevated temperatures . H U F 33040 up to 70 -C. and HUF 22110 up to 120' C . EXFERL%MEK? At .
Membranes were cast in meter-lung tubes .. 2 .8 cm in diameter. quenched in water and assembled for testing in two test units : (a) a laboratory pilot plant in which 3-meter-long tubular modules are assembled and tests made under various conditions in order to determine membrane properties. (b) a trailer-mounted mobile pilot plant designed to perform field tests on a variety of sewage effluents . Kith 2-meter-long membrane tubes_ The luboraiori pilot plan . In the laboratory pilot plant the feed is pressurized by a high-pressure positive displacement piston pomp and high-feed velocities are achiesed by use of a centrifugal recirculation pump which recirculates part of the brine . Feed velocities can be varied between 1-4 .75 mss. and pressures between 2-12 atm . The system works in a closed circuit, i.e. brine and product are recombined
HYBRID REVERSE C64
IS - L'LTRAFRLTRATlO3`\ MEMBRANE
in the feed tank. In order to present oserheating of the system- the recirculated brine passes through a cooling system. Ten 3-meter-long tubes can be mounted in the laboratory plant_ Membrane performance is . time_ linear velocity. pressure and fwd concentration . as +seli as rejection of three different salts - Na,SO .,. NaNO, and KH PO,4 -- xere determined . Na,SO ;, and NaN03 concentrations -Acre determined -cord tuctornetricaily, and P0 4 concentration by gravimetric or e :formance spectrophotornetric methods Dependence of e on linear selocit% was tested by operating the plant at the highest linear velocity for one hour. followed bti sampling . After sampling at the highest salue . the linear selocity was los+ered samples for flux and rejection determinations and after one hour at the nex salues were again taken . This procedure was repeated until the loAest value was reached . after which the linear sciocity Was slowly increased in the reser a order until the highest .clue was again attained . For the pressure dependence dete •- nination the membrane was precompacied for 24 h at 10 at. after which pressure was decreased and samples taken after one hour at each pressure . For the determination of concentration dependence of performance . flux and rejection were also measured after one hour at each concentration . At the conclusion of each test with a different solute. the s}stem v.as flushed v .ith distilled txater to ensure complete remosal of the salt-
Mobile SC'F pilot plant The mobile SUF pilot plant l Fig . i i v.as constructed to test the hybrid
Fig . I . Mobile SUF pilot plant .
S. It . SACHS, E . ZISNER AND G_ IIFRSCOVICI
membranes under field conditions . The size of the unit was determined by several factors : the acceptable pressure drop through the unit . the minimum number of tubes required to yield statistically meaningful results, and the physical size consistent with a trailer-mounted system . The heart of the plant is the set of HUF tubular membrane units which are 28 mm in inside diameter and 2 meters in length, contained within perforated aluminum tubes . Each tube contains 1/6 m 2 of membrane . The tubes are connected in series by U-bends, which are designed to have a minimal pressure drop at the linear velocities employed . The unit contains two parallel banks of 10 tubes each . Each bank can be operated separately, or both banks simultaneously . The sampling of the product is possible from each tube independently . The unit was designed to incorporate considerable flexibility in the parameters considered to be important, i .e. feed pressure. recirculation rate and product recovery . Provisions were mace for a variety of pretreatment and cleaning procedures . The cleaning procedures included for testing were the following: (a) Mechanical cleaning, in which a soft polyurethane foam ball, slightly larger oduced into the system and than the inside diameter of the tubes, is forced through under pressure (2) ; (b) Chemical cleaning, consisting of a combination of cleaning agents such as a complexing agent (e .g. EDTA) and a surfactant (e.g. Triton) : (c) Hydraulic cleaning, in %% hich the system pressure is reduced to atmospheric and after a period of time clean water is passed through at high linear velocities and low pressure. The unit is provided with a primary pump used to pressurize the system and a secondary recirculation pump used to produce a high feed velocity over the surface of the membrane . The nominal working pressure and circulation veloc used so far have been 8 at and 3 to/s. respectively . In addition, the unit has a cooling system in the recirculation loop sufficient to remove the heat generated by the recirculation pump. MEMBRANE PROPERTIES
The HUF membranes developed at IDE are of two basic groups - HUF 22110 and HUF 33040 . Two different polymers are used in the preparation of these membranes and solvent systems as well as additives differ also . These membranes carry a fixed charge capacity which is easily varied during manufacture . Salt rejection and water permeation are affected because of this fixed charge capacity. and therefore these variables are easily tailored to meet specific requirements . Typical performance of an HUF 22110 and an HUF 33040 are given in Table I . As can be seen from the table, flux values for both membranes arc quite
HYBRID REVERSE OSMOSIS - ULTRAFILTRATION MEMLZRANFS
TABLE I PERFORMANCE OF HYBRID MEMBRANES AT SALTS
ATM AND A LINEAR VELOCITY OF
4.7 M/S WITH VARIOUS
NaNOa 200 ppin -- - Na2SO.s 21X} plan ---
H2P04 200 ppm
Permeation Rejection nI 3/rt12 day
Permeation Rejection In'/n2 da} °a
Perneation Rejection m'/nr' ctup
7 .5 8.2
22.5 72 .0
61 .7 90.0
go . a
0 v V
6 7 AP atm
Fig. 2. Performance of HUF 33040 is . pressure.
close, but salt rejections differ appreciably . HUF-33040 exhibits higher salt rejection for all three salts investigated so far . Flux decline slopes for both membranes have been measured with 2000 ppm Na 2 SO4 atm 6 at and a linear velocity of 4.75 m/s. The values obtained are -0 .078 for HUF 22110 and -0.063 for HUE 33040. Performance dependence on pressure, feed concentration and linear velocity were determined for HUF 33040 only . Flux increases linearly with increase in applied pressure (Fig . 2) in the range of 1 atm to 10 atm, while rejection decreases, indicating that the membrane is
S. B . SCM-IS . E_ ZISNER AND G. HERSCOVICI
b Na 3
HUT- - 33040 PR£SSCR£ - S aim
TEM04RATURE - 3OeC LINEAR VELOCITY - 475 inset
4ONSEA. ? 4flQ
Fig. 3. Dependence of rejection on solute concentration for HUF 33040 .
only partially semipermeable, and pressure increase causes salt flow through membrane pores. The rejection of solutes by the HUF membranes must be viewed from two aspects: (a) different types of solutes (b) solute concentration dependence . The rejections of various solutes are shown in Table I, where the dependence on ionic species is clearly illustrated by increasing values in the order NO 3 < SO, < P04 . Decrease in rejection with increase in solute concentration is shown in Fig . 3 . These two properties are qualitatively in agreement with the ion-exclusion principle expected for a charged membrane . At low concentration, log (I - R) vs. the logarithm of concentration (Fig . 3) should approach a slope of I for NaCl and 2 for Na,SO .;, if behaviour were ideal, that is, flux high enough to realize m resection, and flow of solvent and solute completely coupled_ These slopes are clearly not reached, but it would be too much to expect ideal behaviour, and a quantitative test would require estimation of the effects of activity coefficient variations . We attempted to take non-ideality into consideration, but the fit was poor. It seems that factors other than ion exclusion also affect rejection here ; however, ion exclusion is clearly a dominant factor .
HYBRID REVERSE OSMOSIS - ULTRAFILTRATION MEMBRANES
Performance dependence on linear rclocity Decrease of production rate with time is one of the major problems encountered when membrane processes are applied to waters with a moderate or high content of suspended solids . This problem is enhanced when membranes exhibiting high fluxes and high solute rejections are used, owing to concentration polarization . A commonly-used solution to this problem is operation at high linear velocities . However, this leads to higher energy consumptions and, when evaluating the economic feasibility of the process, higher production rates will have to be weighed against the energy costs required to maintain them . In order to determine the appropriate linear feed velocity to be used in the first field tests of the hybrid membranes, performance at various linear velocities was determined in the laboratory pilot plant with a 200 ppm Na 2 SO .~ solution at 6 atm . Results are given in Fig. 4. As can be seen in the figure, fluxes remain fairly constant, but rejection decreases steadily with decreasing linear velocity . When the linear velocity was gradually increased after the lowest value was reached, previous values were again attained, indicating that the decrease in rejection %k as due to increased concentration polarization at lower linear velo-
BLF -330YO AP - b atrns Temperature - 30°C Na2504 - 200 ppm
© - Decreasing Linear Velocity 0 - Increasing Linear Veloc .ty LO i 1 i I 1 i 1 i
30 25 20 Linear Velocity (m/sec)
'*0 1 Q r
n ,, o °,
if 10 `-
c 1 , i
Linear Velocity (m/sec) Fig. 4. Membrane performance vs. linear velocity.
S . B . SACHS, E . ZISNER AND G . HERSCOVICI
TABLE II CONCENTRATION POLARIZATION IN AN "UK 33040 MEMBRANE AT VARIOUS LINEAR FEED VELOCITIES - WITH 200 prim Na2SOI AT 6 ATM AND 30'C . Linear Velocity tnjs
Permeation In "'IY12 f/11.1
4_72 3 .35 1 .10
6.3 6 .5 6 3
67 65 42
Concentration polarization Cis ICu 1 .36 1 .51 2 .33
cities . Concentration polarization values for the system were calculated according to Loeb and Rosenfeld (3) and results given in Table 11 . The effect of decreasing linear velocity on concentration polarization is clearly seen here . As a result of this experiment it was-decided to run the first field test at a linear velocity of 3 nits .
The first field tests were performed at the -Dan Region Wastewater Treatment and Reclamation Plant" (Dan Plant) . This installation treats at present (1975) between 40000 to 70000 m3/day of waste water, and its final stage (1985) will handle the effluent of a population of about 1 .300,000 inhabitants, in a flow of 160 million m3 per annum . The biological treatment of the waste water in the Dan Plant takes place in oxidation ponds built in a system of "first", "second" and "third" ponds operated as recirculation ponds (4) . Raw sewage enters the primary ponds . and is mixed at the entrance with recycled effluents from the third ponds . The effluent of the third ponds was used as feed to the SUF Pilot Plant, installed in June 1975 . The ranee of values for the main contaminants measured in the combined wastewater flow in the period of May to October 1975, and composition of oxidation pond effluent fed to the SUF plant is given in Table Ill .
TABLE III COMPOSITION OF RAW SEWAGE AND OXIDATION' POND EFFLUENT AT THE DAN PLANT Sewage constituents
Raw sessage mg/l
Oxidation pond effluent "Will
Total BOD Total COD Suspended solids (SS) Total nitrogen Ammonia nitrogen Total P
40-70 320-400 100-160 30-40 8-26 13
500-600 300-500 100-120 60-80 12-16
HYBRID REVERSE OSMOSIS - ULTRAFILTRATION MEMBRANES
HUF - 33040 Average Pressure ° 76 atm Temp 35°C
-Sponge Ball Clsanrnq
t- Tap Water FIushrnq
L . It . W
's coor=-.ose =
t I t
t I I I I I I I I I I I I$
Fig . 5 . SUF pilot plant permeation history with oxidation pond effluent .
The first set of experiments with the SU F pilot plant at the Dan Plant site was carried out on 10 tubular HUF 22110 membranes, without any pretreatment of the oxidation pond effluent entering the unit . In addition, no special care was taken to maintain a constant water permeation . by systematic use of one of the suggested cleaning procedures . Initial flux was 4 .3 m 3/m 2 day but after 70 h the flux declined to 50",', of the initial value . At this point of the run a simple mechanical sponge ball cleaning restored the water flux quite effectively, but if too much time was allowed to elapse between cleanings, flux decline was considerably worsened and a more rigorous cleaning was needed . Chemical cleaning did help restore part of the flux but after six hundred hours, cleaning procedures could not restore fluxes effectively and HUF 22110 membranes were replaced with a set of HUF 33040, Fig. 5 . Here, too, the pond effluent pumped into the plant received no pretreatment whatsoever. However, the mechanical cleaning procedure with the polyurethane foam ball was applied once every twenty-four hours . This treatment was performed during system operation and did not require plant shut-down . In addition, tap water flushings for short periods were interspersed between the mechanical cleanings . As may be seen from Fig . 5. on the permeation versus time plot, the water flux restoration obtained as a result of mechanical cleaning is significantly higher than for the flushing operation . In order to evaluate the long term membrane
3 .50 10 .65
7 .31 713 .60
N y z m
Y n x
Dan o.uthttion poor!
32 .34 21 93
SUP pilot plant
. i lion
Secant h a )'
prim, to KU tteaunrnt (3)
Hemu't-Sun Jacinto Valley Secondary EtIluenr
COMPARISON III I WE) N I I •I
tA)I N rs I ROM HO AND SUI PILOT PLAN rs
HYBRID REVERSE OSMOSIS - ULTRAFILTRATION MEMBRANES
TABLE V BACTERIAL ANALtSIS Etttuenrs
Raw Oxidation pond SUF Product Tap water
Coil /100 cc 2 .1 4 .6 ' 23 4
Coil fecatis r
1 .8 . 10'", 0 0
7 4 .3 0 0
. 10-• 10 ;
flux, we have also plotted on a log-log scale, the membrane permeability constant as a function of time . The -values for these points were calculated from the permeation data immediately prior to mechanical cleaning . Thus, the plot presents a higher-than-average flux decline . This is a more accurate representation of membrane performance on oxidation pond feed, because of the short-term improvement achieved by each cleaning . Hossever, Nshile the immediate improvement is short-lived due to the high concentration of particulates present in the feed, the long-term flux is maintained at a much higher level than would be the case with less regular cleaning (for the HUF 22110 membranes operated with irregular cleanings . the flux decline slope was -0 .32 as compared to -0 .098 for HUF 33040) . The quality of the SUF pilot plant product is given in Table IV . As may be seen, the reduction in ROD is 93 dissolved COD is reduced by 80% and total COD is reduced by 94 .6" . . It should be noted that the BUD values of the product have reached the experimental limits of the measurements . The total dissolved solids, which are mostly inorganic in nature, are also reduced by 19%, which is not yet at the 30 ° . reduction target level . As expected for this process, the reduction in suspended solids and turbidity is almost complete from 100 JTU to 0.30 JTU . A 48 °„ reduction was achieved for NAB, while that of ortho-phosphate was 88% . Samples were taken from time to time for bacteriological analysis . the results of which are shown in Table V .
SUMMARY AND CONCLUSIONS
The first series of experiments performed at the Dan Plant with the mobile sewage treatment pilot plant containing a new class of hybrid ultrafiltration membranes has been described . The performance of these membranes may be considered as an intermediate between a charged reverse osmosis membrane and a conventional ultraflter. It exhibits high water fluxes at moderate pressures, as compared to reverse osmosis membranes and rejects a moderate amount of certain salts, particularly multivalent ions . In addition, these membranes have demonstrated a marked capability for reducing oxidation pond effluent contaminants to a
S . It . SkCIIS . E. ZISNER AND G . IIERSCOVICI
6 To: at COO
Fig . 6 . Comparison of effluent qualities from various treatments .
lower level than other existing biological or physico-chemical processes operating under similar conditions . Fig . 6 gives a comparison of effluent qualities from the various treatments, including those of an HUF membrane and a "sludge blanketsolid contact-lime reactor clarifier" . By conventional water quality standards . the HUF membrane yields product water which can be used almost without restrictions in agriculture, industry or for groundwater recharge . Comparing these results with those of the lime reactor clarifier one sees that the HUF membrane is much more effective in removing the major waste water contaminants such as BOD . COD and suspended solids . It is also of interest to compare the present results (Table IV) with an EPA study (1) of RO treatment of Hemet-San Jacinto Valley secondary effluent . This pilot study was carried out with conventional RO equipment from four different
HYBRID RIA ERSE OSMOSIS - UL rRAFILTRATION MEMBRANES
manufacturers to determine its feasibility on untreated and treated secondary effluents . Rapid membrane fouling occurred in spite of the use of post-secondary effluent treatments such as alum clarification, sand filtration, granular activated carbon- chlorine addition and pH adjustment . and combinations thereof . The flux decline slope of the most successful and longest experiment' as -0.08!, which is very similar to the slope of -0 .09 obtained with the HUF membrane. However, it should be noted that the former was obtained with all three listed treatments plus pH control . Furthermore, the RO membrane had an average membrane constant, A . 1/20 of that of the HUF membrane, which operated on untreated oxidation pond secondary effluent . In conclusion, except for the reduction of TDS . there is no real advantage in the use of conventional RO units in sewage treatments . ACKNOw'LEIK,E31EN FS
The authors wish to thank Prof. G . Stein of the Hebrew University fo his encouragement and constant guidance . Dr. Jerry Tanny of the Weizmann Institute for his helpful suggestsons . and Mrs. Sarah Cohen, Mr . Menashe Cohen and Mr . Tibor Meir for their %aluable technical help . REFERENCES
D . F. BOEN AND G . L. JotnNNSFN, Reverse osmosis of treated and untreated secondary sewage effluent. EPA 670;2-7d-077, September 1974. 1 S. LOEB a'D E. SILLOVEk . 16 months of field experience with the Coalinga plant, Desalination . 2(1967) 75-80. 3. S. LoEB AND J . Rost-FEt0 . Turbulence region performance of RO desalination tubes . Department of Engineering . UCLA Rept No. 66-62. October 1966 . 1.
4. G . SUELEF, M. Ro\EN AND M . KREMFR. Waste water recirculation - pilot plant and field studies, submitted for presentation at the 8th Intern . Conf. on Water Pollution Research . Sydne) . 1975.