As water and sewer costs are expected to increase over the next decade, membrane technology will continue to be a viable and economical option in the recovery and reuse of water. This paper describes the mechanism of membrane effected separation (specifically reverse osmosis, ultrafiltration and microfiltration), membranes commercially available, membrane element configurations, and how complete membrane systems are designed. The paper also discusses case histories of a variety of proven applications, as well as areas of high potential. System design, operating parameters and detailed benefits derived from use of membrane systems are also included.
Laboratory, pilot plant and full-scale unit process equipment is presented, as well as the steps required to evaluate feasibility and scale-up to process conditions. Economical evaluation of the process is presented in the case histories. New areas of high potential for payback on water reuse and/or recovery of valuable products, as an added benefit, are also described.
Although a relatively young technology, long term success has been achieved for water recovery in many areas, including process applications. Improvements in equipment and membrane are reducing the cost of these unit processes. Membrane technology will play a major role in all areas of water recovery in the near future.
Crossflow Membrane Filtration
Membrane filtration is the separation of the components of a pressurized fluid performed by polymeric membranes. The openings in the membrane matrices (pores) are so small that significant fluid pressure is required to drive liquid through them; the pressure required varies depending on the size of the pores. Reverse osmosis (RO) membranes have the smallest pores, while microfiltration (MF) membranes have the largest pores, and hence, require the least pressure.
Normal particle filtration has historically not been run in a crossflow design, 'perpendicular flow' may be the most appropriate term, with the solution to be filtered approaching the filter media in a perpendicular direction. The entire influent stream passes through the filter media (except for the particulate matter filtered out). Filter media can be either of the 'depth' type or the 'surface' or 'barrier' type. In this perpendicular flow design, there are only two streams, the influent and the effluent. separation is effected in the micron range or greater; with certain depth filter media achieving as low as a nominal one micron separation.
Crossflow membrane filtration is fundamentally different in design, in that the influent stream is separated into two effluent streams, known as the permeate and concentrate. Permeate is that fraction which has passed through the 'semi-permeable' membrane. The concentrate is that stream which has been enriched in the solutes or suspended solids, which have not passed through the membrane. Membrane is a surface filtration media, which effects separation in the ionic and molecular range, as well as the macromolecular and particle range. The advantage of this design approach is that the membrane media is operated in a continuously self-cleaning mode, with solutes and solids swept away by the concentrate stream, which is running parallel to the membrane. Hence, the term 'crossflow', and the advantage of this approach for separation in the micron, sub-micron, molecular and ionic range.
Figure 1 shows an artist's conceptual view of crossflow membrane filtration. The feed solution flows under pressure between the two membranes depicted, and as it flows over the membranes, permeate passes through. The rate of permeation is known as flux. The concentrated fraction of the stream exits through the same flow channel as the feed enters, carrying away concentrated solutes and particles.
Mechanism of Separation
Figure 2 shows the mechanism of RO. Pores in this membrane class are so small they have not been resolved even by scanning electron microscopy. The pore diameter is theorized to be somewhat smaller than twice the thickness of the pure water layer, which occurs over the membrane (1). This pure water layer is void of ions, and through this exclusion 'rejects' the ions, preventing them from permeating the membrane. Note that very small molecular weight organics and uncharged solutes may pass through the membrane along with the solvent (usually water).
Reverse osmosis, sometimes called hyperfiltration, is perhaps the most technically complex class of membranes, effecting separation at the micromolecular and ionic size range. Pore sizes ranging from 5 to 15 angstroms effect separation of the solutes down to 150 molecular weight and often lower. As with ultrafiltration (UF), RO has rarely been viable unless in a crossflow mode.
Ultrafiltration is perhaps easier to understand, and as Figure 3 shows, the mechanism of separation is largely based on size (sieving). The pores are too large to effect desalting, and ions readily pass through. For the UF membrane shown, organic molecules under 1000 molecular weight may permeate the membrane. It should be noted that all discussions where molecular weights are used as an indicator of size assume that the molecule is a polysaccharide of that molecular weight. Ultrafiltration is generally defined as effecting separation in the 0.002 to 0.2 micron range. This is perhaps more usefully described as the 500 to 300,000 molecular weight cut-off (MWCO) range, requiring pore sizes of from 15 to 1000 angstroms.
Microfiltration membranes in the crossflow mode work on the same principle as UF, although materials small enough to be present in dissolved form will generally permeate the MF membrane while suspended solids are rejected. Microfiltration has generally come to be regarded as effecting separation in the 0.02-2.0 micron range. Microfiltration has historically been run in the perpendicular flow mode, requiring disposal of the membrane media due to blinding by the retained material. Interest in MF crossflow has been increasing lately and development of the use of this membrane class in various crossflow configurations is also increasing.
Membrane Polymers Commercially Available
The wide range of water recovery applications comprise a variety of fluid environments, some of which may be detrimental to certain polymeric materials and the morphological characteristics of membrane. Operating conditions such as temperature, pressure, solution pH and chemical compatibility must therefore always be considered. No one polymer will withstand the environments of all the possible water recovery applications where crossflow membrane processing can be applied, so several different membrane materials should be investigated to select the optimum for each situation.
Reverse Osmosis polymers currently available include the asymmetric membranes: cellulose acetate (CA), aromatic polyamides (PA-type) and a patented polymer from Osmonics, which is also a PA-type. Also available are the newer thin-film composite membranes (TFC), of which aromatic polyamide (PA) on a polysulfone substrate is the best known.
Ultrafiltration polymers currently available include polysulfone (PS), CA-blends, a fluorinated polymer (VF), and Osmonics' patented polymer (PA-type).
Microfiltration membrane polymers commercially available are the most numerous, due to the relative ease of achieving the required pore size for MF compared to UF and RO membranes. Polypropylene, acrylonitrile, nylon and PTFE are among the more common MF polymers, and their broad chemical compatibility characteristics make the MF range of membranes the most chemically stable.
Developmental membrane polymers that show some long-term promise are acrylonitrile copolymers and polyethersulfone, although neither is beyond the laboratory stage yet.
In choosing a polymer membrane, the following environmental conditions must be evaluated:
Microfiltration membranes are typically operated at pressures from 0.2 to 3.4 bar (3 to 50 psig). Ultrafiltration membranes from 1.03 to 13.8 bar (15 to 200 psig) and RO membranes from 13.8 to 69.0 bar (200 to 1000 psig). Under the higher operating pressure conditions, membranes may be subject to mechanical compression (compaction) and other mechanical deformation, which can alter performance over time. The result of compaction is irreversible loss of permeate rate (flux), and compaction has long been identified and observed in the RO industry. However, recent experimental evidence suggests that irreversible fouling is more of a factor in flux decline than true membrane compaction (2). Thus, understanding the membrane's characteristics and selecting the most appropriate membrane for the pressure required is an important consideration.
Since some processes in industry are performed at elevated temperatures, careful consideration of polymer temperature limits is essential. Crossflow membrane process separations are currently operating at a temperature range from 0 to 85°C (32 to 185°F). Membrane cleaning and steam sterilization of the system can occur up to 110°C and higher. To further save process energy costs, it may often be preferable to operate the membrane system at elevated temperatures due to pasteurization or other processing requirements.
The best understood and least expensive RO membrane, CA, has often had an upper temperature limit assigned to it in the 35 to 40°C (55 to 104°F) ranges. This limit appears to have no scientific basis and was probably set because of membrane element restrictions rather than the limits of the membrane itself. Recent data from controlled experiments show that CA membrane retains its separation performance quite well at much higher temperatures and is effective in processes to 65°C (145°F). CA does not appear to be steam sterilizeable but additional research work is continuing. PA type RO membranes definitely have higher temperature limits and can operate successfully at 82°C (180°F) at 600 psig and even higher temperatures under lower operating pressure.
Ultrafiltration and microfiltration membranes, operated under much lower pressures and typically made of more temperature resistant polymers, can often be operated up to 100°C. Under these conditions, it is not unusual for the other membrane element components to be the limiting factors, not the membrane itself.
Although pH is often considered in the context of the overall solution chemistry, the degree of acidity or basicity itself also affects membrane life. Basic solutions have historically been a problem for CA membranes since the rate of hydrolysis increases exponentially with increasing pH values. The generally accepted pH range for CA membrane is 2 to 8, although higher and lower pH values are feasible if the economics of the application allow more frequent membrane replacement, or if ionic rejection is not critical to the application. PA type RO membranes are generally rated as compatible up to pH 11 or 12, yet many on the market are not rated below pH 3 or 4 in the acid range. Among the UF membranes, pH range tolerances are often greater, with PS membrane frequently rated as compatible from 0.5 to 13. MF membranes are generally even more pH tolerant. The limitation of pH for the more sensitive RO membranes is the major reason for the push to find new 'non-cellulosic' RO membranes.
Along with pH range, general chemical compatibility is also a major requirement for choosing a membrane polymer. Lack of chlorine resistance has often been a major obstacle to the application of TFC-PA type membranes. Attack of PS membrane by many types of hydrocarbons has resulted in CA membrane being chosen in many water recovery UF applications, despite its lower hydrolytic stability. A frequent limiting factor for MF membranes is extreme hydrophobicity, resulting in the requirement for chemical additives to lower surface tension or operating at much higher hydraulic pressures.
When chemical compatibility limitations are combined with the fact that the most desirable polymers can only be manufactured into certain types of membrane (with limited range of pore size, morphology and production rate or flux), the requirement for a wide array of membranes becomes even more apparent. A wide variety of available membranes, an understanding of the capabilities of each specific membrane, and the continual search for new membrane polymers and membrane morphology are all important factors in applying crossflow membrane technology.
Element Configurations Commercially Available
As the membrane industry has grown and developed, several different configurations for supporting and containing the membrane were developed. With UF and MF membranes, the relatively low operating pressures allow some different configurations than with RO. Commercially available configurations for RO and UF include tubular, larger internal-flow hollow fiber or 'spaghetti bundle', plate and frame, and spiral-wound.
Tubular UF units are generally produced with inside diameters of 12.5 or 25 mm (0.5 or 1 inch) and in lengths of 150 to 610 cm (5 to 20 feet). The feed solution flows through the inside of these tubes, whose inside walls contain the membrane. Although relatively uncommon, reverse osmosis membrane is also available in the tubular configuration. As with ultrafiltration, the feed and concentrate streams flow through the inside of the tube, usually available in 12.5 mm diameters (0.5 inch).
Hollow 'fat' fiber membranes can be thought of as self-supporting, miniature tubular units, with inside diameters of the fibers ranging from 0.5 to 1.1 mm (0.020 to 0.0143 inches). Like the tubular units, the feed stream flows through the inside of the fibers and the permeate is collected on the outside.
The hollow fiber configuration with the pressurized flow in the bore is not used in RO, but a configuration with the pressurized flow on the outside is widely applied. The extremely small size of RO fibers results in their description as 'hollow fine fibers' (HFF). These fibers are much smaller than their ultrafiltration counterpart, having the diameter of human hair (42 microns I.D., 85 microns O.D.). This small size allows an extremely high element packing density, but also makes these elements highly susceptible to fouling, especially in waste treatment applications.
The 'plate and frame' configuration is a third style for housing membrane. Flat sheet pieces of membrane are placed between plates, which form flow channels over the membrane with heights of approximately 0.5 to 1.0 mm (0.019 to 0.039 inches) for UF and from 0.3 to 0.6 mm (0.012 to 0.023 inches) for RO. These plates are stacked in parallel groups, connected with extensive sealing systems and interconnecting plumbing. This configuration is similar to a conventional plate and frame filter press.
The 'spiral-wound' element also uses economical flat sheet membrane; when wound around a central permeate collection tube and separated by thin spacer materials, this design yields a high membrane packing density (3). Specific packing density depends on the size of spacer material used. Spacer thickness can be adjusted depending on the application. For UF, they are generally from 0.75 to 1.55 mm (0.030 to 0.060. inches) in thickness or channel height. The feed passes over the membrane through the spacer material, which also acts as a turbulence promoter to keep the membrane clean at relatively low velocities. Reverse osmosis elements of the spiral-wound configuration are also of the same basic design as with spiral UF. As in the plate and frame design, the flow channel height is generally smaller than in UF, ranging from 0.63 to 0.86 mm (0.025 to 0.034 inches). This results in increased packing density compared to UF (approximately +30%). The fluid flow dynamics are similar in both the plate and frame and spiral-wound configuration.
A recent innovation for spiral elements is the use of channelized spacer material. This new spacer material creates channels with much the same fluid dynamics as small tubes, combining the high membrane area of the spiral-wound configuration with the fouling resistance of the tubular design.
Crossflow MF membrane elements have been available commercially on a limited basis and are the most recent entry into the crossflow membrane field. These have typically been modified pleated cartridge filters and have been applied largely in the pharmaceutical and medical industries where high cost, low volume solutions have been processed. However, spiral-wound MF membrane elements have been in the developmental stage since early 1983. Both in-house application tests and field test results have shown promise for spiral-wound MF use in clarification, color removal and macromolecular fractionation applications (4).
The important differences between these configurations, which must be considered, include the pumping energy costs to maintain efficient operation, pretreatment requirements, cost of replacement membrane, labor and the capital cost of the equipment. (The scope and length limitations of this paper do not allow further discussion of membrane element design. However, a paper presented to the Technical Association of the Pulp and Paper Industry 1983 International Dissolving and Specialty Pulps Conference by the same authors addresses this subject more fully, listing advantages and disadvantages, and comparing the membrane packing density of commercial elements.)
Equipment and System Design
Quality systems engineering is as important as membrane and membrane element selection in providing the process and plant engineer with effective long term membrane system performance. Membrane systems manufacturers should be able to recommend and supply all necessary pretreatment equipment, machine-mounted controls and monitors, Clean-In-Place (CIP) systems and cleaning procedures. A crossflow membrane systems supplier who understands fluid chemistry, fluid dynamics and corrosion factors, who understands the capabilities of various prefiltration techniques can supply a system with the highest chance of performing efficiently and as required. Supplying a crossflow membrane system that is the most effective in both capital and operating cost, as well as in meeting customer objectives for water recovery with long term system performance, is the goal of both the customer and systems supplier.
Machines and systems for industrial water reuse will vary in size and complexity with the requirements of the specific installation. While many water recovery machines will be the size of the large unit pictured in Figure 5 (which shows the 'largest to the smallest' typical membrane units), not all water reuse applications need be on a large scale to be economical. Two of the case histories discussed in this paper on water reuse are augmented economically by the recovery of valuable materials and reduced or eliminated disposal costs; these applications require equipment quite modest in size.
Larger crossflow membrane systems will typically be made of multiple skid-mounted units similar to the one pictured (Figure 5), often with centralized pumps and control panels.
Depending on the application, a certain amount of pretreatment is also necessary to maintain successful operation of the membrane. Membrane equipment manufacturers often supply all the necessary pre and post-treatment equipment. Figure 6 shows a bag filter system for pretreatment to an RO machine, which is concentrating oily waste and recycling pure water. There are several housings plumbed in parallel, which contain 5 micron bag filters in this case (equivalent to approximately 15 micron rated cartridge filters). These filters remove suspended material, which could foul membrane.
Aside from prefiltration, several other controls are necessary or desirable on crossflow membrane equipment. These include, but are not limited to; pH control/monitoring, temperature control, biocidal agent addition, dispersants to prevent precipitation or solute agglomeration, flow meters, pressure gauges, and sensor-activated alarms to alert the operator or to shut the system down when design operating parameters are exceeded.
Areas for Water Recovery
Applications for membrane technology fall into three general categories: process, waste treatment and water purification (generally naturally occurring water sources). Water reuse via membrane separation would generally be expected to come from waste treatment applications where disposal cost or regulations apply, although many of these evolve into process applications with the recovery and reuse of valuable materials. In some cases of high purity or high volume water usage, water itself is a valuable enough resource to justify the process economically.
Most industries use substantial quantities of water, and therefore, would benefit from water recovery. In the metal finishing industry, thousands of plating operations are candidates for rinse water recovery and reuse of plating salts. Combined plating waste streams can be treated successfully, and large scale (50 to 100 gpm) zero discharge systems have been installed for some time (5). Recycling both lubricant oils and rinse water in can-forming and other metalworking applications has also been accomplished in several instances.
The dilute waste streams, which result in the chemical processing industry, can usually be concentrated economically with crossflow membrane technology and process water conserved. As large users of water, both the pulp and paper and textile industries have a vested interest in all feasible water recovery/pollution abatement technologies.
Expensive, high purity water is a necessity for manufacturing miniature electronic components. The large volumes of spent ultrapure water can be purified for reuse by both RO and UF systems (6). The photographic and printing industries also generate large volumes of contaminated water, and can solve their discharge problems while simultaneously reducing make-up water expense by recovery for reuse (7).
The advances in membranes, membrane elements and attendant hardware discussed earlier will allow the recovery of not only water, but precious energy in the form of heat. Boiler blowdown and cooling tower blowdown streams are prime candidates for this type of membrane application.
Following are four specific case histories where crossflow membrane has successfully and economically been applied for water and resource recovery.
Oily Waste Treatment and Water Recovery
A die cast manufacturer for automotive parts, Sealed Power Corporation, was discharging their treated end-of-pipe process water to a lagoon. Sulfur compounds and their associated smells created a problem for a small community located nearby. Sealed Power chose to evaluate alternative forms of disposal compared to conventional flocculation and lagoon storage. The location of the nearest mainline for city sanitary sewer (5 miles) made it uneconomical to consider sending the process waste water to the city treatment plant. Estimated cost to tie into the city sanitary sewer was $1,500,000 (8).
Osmonics was contacted to evaluate the use of an RO system to concentrate the treated water after flocculation to minimize hauling costs and reuse the permeate in the cooling towers. A feasibility application test was performed on a representative sample to choose the optimum RO membrane. Results from an application test membrane scan showed the membrane of choice was
Sepa® -92(CA). Next, a pilot unit was installed on-site to gather long-term operating data. Severe membrane fouling occurred due to the variability of the feed stream in terms of oil content and TDS. The conventional flocculation, functioning as pretreatment to the RO, used both caustic and acid addition, which added considerable TDS to the treated wastewater. Following four months of pilot operation, new RO sepralators were installed and UF was added as pretreatment to increase overall membrane efficiency. This system was then compared to conventional flocculation as pretreatment ahead of the RO.
Testing was performed on the UF/RO system to generate long-term data from which to design full-scale equipment. After evaluating results from both methods, the clear decision was to proceed with UF/RO based on economics and the operating problems associated with flocculation.
The annual cost to treat the process wastewater with conventional flocculation was $60,000/year in materials and 20 man-hours/month for batch operation. In addition to these costs, the plant was severely limited in choice of die lubricants that were compatible with the chemical treatment system. This limitation had an estimated cost of $l5,000/month due to reduced performance and higher lubricant cost (9).
Make-up water to the die cast operation was well water with a TDS of 450 ppm as CaCO3. The dissolved solids in the well water shortened life and placed an additional load on the high recovery waste treatment RO. The decision was made prior to installation of the waste UF/RO system to install a separate RO system to purify 4 gpm of make-up water to the die cast operation, reducing cost by increasing overall system efficiency. The schematic diagram in Figure 7 represents the final installation.
The total installed cost of the UF/RO waste treatment system was $200,000. Yearly operating cost is $95,000, which includes replacement membrane, chemicals, electricity, labor and hauling of the concentrated waste. The overall system recovers 5000 gpd of water for reuse, saves $136,600/year on hauling costs and saves about $180,000/year materials cost on the flexibility to purchase a wider variety of die lubricants.
Recovery and Reuse of Spent Electronics Grade Rinse Water
A major semi-conductor manufacturer recycles 1100 gpm of DI water 24 hours/day. This recycled water was previously discharged to the drain after rinsing electronic components.
Employing RO, the spent DI water is now reclaimed for reuse. separate rinse streams are combined with make-up DI water and fed to the RO system with a quality in resistance of 3 to 5 megohm. The RO systems operate at 90% recovery (recovery is the fraction of feed stream volume produced as permeate, expressed as a percentage) and the RO permeate is fed to polishing DI tanks through 0.2 micron filters and back into the DI loop, now with the extremely high quality of 18 megohms. Over 80% of the spent DI water, which was formerly discharged to sewer, is now recycled on a continuous basis. Figure 8 is a schematic representation of the water recovery and reuse system.
The economic incentive to recycle spent DI water from an ultrapure water system is considerable. At a large installation in Essex Junction, Vermont, it is reported to cost $20/1000 gallons to produce essential ultrapure water (10). While this cost is high, the extreme microchip scrap rate that results without an ultrapure water rinse is much more costly. On this basis, a system recycling 1100 gpm continuously has cost savings of over $11 million/year.
The major costs of an RO system operating on spent DI water is electricity, followed by membrane replacement. Electrical consumption at this installation is 515 kilowatts and at $0.06/kilowatt-hour, running continuously, the annual cost for electricity is $270,000. The sepralators (membrane elements) have been changed after 36 to 48 months of operation. On a 36-month basis, the annual cost for membrane replacement is $220,000/year.
The total installed cost of the DI water reclamation system, including the polishing Dl tanks and point-of-use 0.2 micron filters was $3,000,000. This showed a payback in less than 100 days.
Nickel Plating Rinse Water Recycle
The use of RO to reclaim nickel salts and recycle water to the rinse tanks is a time-proven application. Over the past 10 years, Osmonics has installed over 140 units for metal plating salts reclamation. Numerous systems had a payback of less than one year.
An example, typical RO system installed on a nickel line in the Midwest processes 120 gph (454 Lph) at a recovery of 95%. The permeate stream (116 gph [439 Lph]) is recycled as rinse water and the concentrate stream (6 gph [23 Lph]) containing the nickel salts is returned to the plating bath. Nickel plating lines usually will operate at elevated temperatures, 120 to 140°F (49 to 60ºC), which causes evaporation from the open plating tank. The RO systems are designed with the concentrate flow equal to the evaporation, which creates a steady level in the plating tank. Make-up pure water is added to the rinse tanks at 6 gph (23 Lph) to counter the loss of water by evaporation. Since it is advantageous to use RO to produce this make-up water to prevent tap water salts build-up in the bath, this was done at this installation. Figure 9 shows a typical flow schematic of a nickel recovery system.
The RO system processing 120 gph (454 Lph) recovers 3.5 lbs./hour of nickel salts at cost savings of about $1/lb. of salts. The plating line operates 16 hours/day, 5 days a week, which correlates to nickel salt savings of $14,000/year. Without an RO system, the rinse water needs to be chemically treated and the sludge disposed of as a hazardous waste. Disposal of hazardous sludges is increasing in complexity and cost, and the EPA plating discharge guidelines which went into effect in April, 1984, have reduced disposal alternatives. Hazardous sludge disposal can generate costs that currently exceed $1/lb. (11). The minimum disposal cost would start at $14,000/year. As can be seen, the major economic advantage of an RO on a nickel line is reuse of the plating salts and recycle of the rinse water which eliminates any discharge.
The installed cost of the nickel reclamation RO system, make-up pure water RO and transfer pump to deliver rinse water to the RO was $27,000. Payback on the equipment was achieved in less than one year in this case. Another nickel recovery system in a large shop cost $16,000 in 1982 and is saving over $50,000/year in plating salts and chemicals alone, and even more when viewing disposal costs saved (12).
Copper Cyanide UF/RO System
A large midwestern plater had installed a cyanide destruct system to prevent cyanide from entering the city sanitary sewer. The destruct system was able to remove the majority of the cyanide, but complex ferricyanides still present in the discharge stream exceeded the city standards. Also, this conventional treatment of the CuCN plating rinse water produced a water quality which was not acceptable for reuse. Osmonics designed and manufactured a combination UF/RO system to recycle rinse water, reclaim plating salts, eliminate any discharge to the sewer, and prevent the handling of toxic sludges for disposal.
The UF/RO system processes 200 gph (757 Lph) of rinse water, producing an RO permeate stream of 190 gph (719 Lph) and an RO concentrate stream of 10 gph (38 Lph), and the system is designed based on Osmonics’ patent 3,637,467. The UF portion is used as pretreatment to the RO to remove the oils and suspended solids typically found in cyanide plating baths, which reduces RO membrane fouling and increases membrane life. Concentrate produced by the UF system is recycled back to the rinse tank. Continuous cartridge filtration to 25 microns on the rinse tanks removes any contaminants that might otherwise build up in the bath. Figure 10 schematically illustrates the flow and components of an installed system.
Total installed cost of the UF/RO system was $30,000. This price includes a pure water make-up RO unit and a transfer pump to deliver feed from the rinse tank to the UF/RO.
As originally installed, the UF/RO system was using a DuPont hollow fiber membrane RO element (permeator), due to the better resistance of its PA membrane compared to the other available RO membrane polymer, CA. Short and uneconomic permeator life occurred due to the inability of the hollow fiber design to tolerate precipitated tap water salts (introduced by design error elsewhere in the plating system). The permeator package materials were also apparently unable to tolerate the pH of the rinse water (pH 9 to 10). Permeators had a useful life of only 2 to 3 months, and after subsequent failures, the permeator was removed and replaced with spiral-wound PA sepralators. After 14 months operation, the sepralator performance data was the same as at start-up, with over 98% rejection of the TDS. Figure 11 shows the CuCN recovery machine prior to replacement of the permeator.
Method of Application Development
For membrane applications other than natural source water purification, there are a series of steps which are helpful in developing the proper membrane system for a particular application. Short-circuiting these steps may result in an underdesigned system, so understanding the steps is important for all potential membrane users. The development steps listed below presuppose an application such as waste treatment, where there are no prior successful tests or production installations on identical fluids. Care must be taken to assure solutions and process conditions are nearly identical before eliminating these steps.
If potential membrane users have their own laboratory, they may wish to undertake in-house investigation of initial membrane performance. Samples of commercial flat sheet membrane and some developmental membranes can be purchased from membrane element manufacturers. Traditional laboratory equipment to run these tests includes the 'stirred cell'. This apparatus can yield a rough idea of the fluid separation possible, but is not a crossflow apparatus and does not provide accurate data on flux rates or the degree of concentration feasible. Far superior for lab testing are crossflow test cells which are available on the market. These devices test small pieces of several different membranes quickly and inexpensively, and with the important crossflow fluid dynamics. A feed, permeate and concentrate stream are all generated in small quantities. Figure 12 shows a stainless steel membrane test cell capable of 20 to 1000 psig (1.4 to 69 bar) operations. Lower pressure, less expensive plastic cells are also available.
To generate data for scale-up to a full-size commercial installation, the best testing procedure is batch processing conducted with Process Evaluation Systems (PES) which are equipped with full instrumentation and all the basic components of production scale equipment. Systems such as that in Figure 13 are available for rent or sale for use at the processor's facilities and are also used for application tests at the membrane supplier's facilities. Such systems can quickly scan numerous membrane elements to determine optimum performance. This approach yields the best short-term data for membrane type, element configuration and general equipment hardware.
The next step is long-term testing with a prototype pilot system at the processor's facility. Pilot plant test equipment, complete with pretreatment and Clean-in-Place hardware, is commercially available to run on batch process solutions, small process streams or slip streams from larger processes. Figure 14 shows a production size PES system complete with two sanitary construction high pressure pumps mounted in series, and full instrumentation and controls to allow monitoring of total system performance. Testing of two to three months, with daily performance monitoring, will provide accurate performance data and the necessary information on membrane/hardware interaction with the solution to allow a membrane systems manufacturer to guarantee the production scale system.
Experienced and qualified membrane system vendors will work closely with potential users to determine the application testing approach best suited to their fluid characteristics and separations objectives. The result should be a full scale crossflow filtration system with guaranteed performance and assurance of long-term successful operation.
In conclusion, membrane technology has come of age, has been field proven for economical water recovery, and therefore, is a viable alternative that should be considered by all process engineers, plant managers, environmental engineers and researchers. Continual improvements in membrane, element construction and system design are expanding the capability of crossflow membrane separation, and all areas of industry can benefit from this technology.