Any new technology must overcome perceptions of the marketplace, which are not accurate. As a technology moves into new areas of application, new misconceptions also arise. Crossflow membrane technology is no exception.
The first man-made, pressure-driven 'crossflow' membranes were made in flat sheet form. This membrane had to be configured in a device, and was first made in tile obvious 'plate-and-frame' design. The spiral-wound design resulted as a response to the high cost, large size, and impractical seals of the plate-and-frame. The necessary use of a mesh feed channel spacer caused a perception that spirals could not process high solids waters. Additionally, tile complicated flow paths through these mesh spacers were not amenable to simple fluid dynamics modeling. A general perception grew that spirals were not applicable for many applications, so both hollow fibers and tubular element configurations were reviewed as preferred for many applications (round, open channels). Yet from about 1975 to 1990, spirals were proven feasible and indeed more favorable for many high solids applications. They were proven in both their original design, and incorporating feed channel design improvements.
The cost-effectiveness of spirals drove their acceptance and they still dominate most process and water treatment applications, including high fructose corn syrup purification, dye and ink desalting, and commodity fermentation product purification. They are even now replacing hollow-fiber elements in ED paint reclamation. Fat hollow fibers and tubular configurations are more capable for some, but few, high solids applications.
Initially, crossflow membrane technology was seen as too expensive for potable water treatment, except in the most extreme cases of need. Man-made organic pollutants drove municipalities, the AWWA, and the EPA to closely evaluate membranes in the 1980's. However, actual installations to remove such pollutants were rare. Beginning in the late 1980's, and accelerating in tile early 1990's, waterborne disease outbreaks and the realization of widespread surface water contamination by protozoa fueled even more intense evaluation. Municipal systems installed for microbe removal are now catching up with those for inorganic and man-made organic contamination removal. In 1997, there are few who doubt membrane's role in treating difficult water sources for wide potable use.
Municipal water treatment operators must assure their unit operations' performance are adequate on an ongoing basis. Accurate testing for microbes is more difficult than most other contaminants. The need to test the membranes, then, in water on a routine basis preferably in-situ, may be seen as a substitute. Most membranes are a barrier of thin polymeric material, and rely on seals within their element configurations and also within their larger systems, the question of assuring integrity is natural. The relative ease with which a particular design can be integrity tested may then be a factor in a purchase decision. Once again, with some potential users (but as shall be seen - not all) the spiral-wound configuration may face a perception problem.
Microbe Removal by Membranes
The potential for failure or substandard performance exists for any technology. Crossflow membrane for potable water has been applied despite this potential. Such use in the United States is in response to increasing EPA mandates for water treatment steps. Most membrane technologists have heard of the Safe Drinking Water Act (SDWA) and those Rules which followed - most notably the Surface Water Treatment Rule (SWTR). The many (and increasing) requirements of those rules will not be stated here. Instead, Table I from Professor James Taylor's referenced paper 'is reproduced as a summary of the need, and the accepted capability of membrane processes to meet the various EPA requirements.
Microbe removal (pathogens) is a key expectation.
That pressure-driven membranes present an inherently effective barrier against waterborne microbes is a given among those who design, study, and use them. Removal efficacy has been proven in countless published and unpublished studies. The questions then are: What is the potential for flaws, which would leak microbes through the large areas of membrane which commercial installations require? Similarly, what is the potential for leaks through the seals within elements and the additional seals within the system? That there is some potential for leaks is understood: How does one detect such leaks?
Spiral VS. Hollow Fiber Configurations
Note that microbes are only one category of contaminant that must be removed, and that membranes have the potential for removing several categories. This fact heavily favors the spiral configuration, as will be discussed later.
The strong and weak points of each configuration have been debated and presented over the last two decades. 2,3 Table 2 highlights those that are key in the opinions of the authors.
Two developments since this comparison was first made should be pointed out:
The practice of 'backflushing' (permeate side to feed side flow) to clean the self-supporting UF and MF hollow fibers was a breakthrough. Also, 'clear channel' and asymmetrical design feed channel spacers have been developed for spiral-wound elements.
The history of crossflow membrane in potable water treatment is useful in the comparison of spirals to hollow fibers. The first use was on seawater, where DuPont's hollow fine fibers (flow from the outside in) initially dominated the market. These were not economical for other applications, however, because their tendency for severe fouling required expensive pretreatment schemes. Then Cadotte's composite flat sheet membrane was commercialized and spirals could treat seawater economically. As membrane was applied to other potable applications (so-called 'brackish' water desalting, color removal, sulfates, and alkalinity removal for taste enhancement, etc.), spirals dominated the new applications.
When trihalomethanes (THM's) and other disinfection by-products (DBP's) became a concern, the various classes of membrane were tested against them, and then their organic 'precursor' molecules. These tests were predominantly done in the spiral-wound configuration. Intense study by Florida municipalities, consulting engineers, and universities soon led to the conclusion that UF-class membranes could not adequately remove precursor molecules; NF-class membrane was required. The need for NF both improved the stature and focus on NF membranes, and precluded use of the fat hollow fibers (although spirals were already used to a large degree). Color removal had earlier been studied and tight UF membrane was often found sufficient, but the concern for precursors DBP's eclipsed the color removal concern, and few UF systems were installed for that purpose.
With the 1990's came increased concern about microbial contamination and with the 1993 cryptosporidium-related deaths in Milwaukee, interest in the U.S. increased in membrane's role in potable water treatment. Where only UF or more likely MF separation was required, the ability to b4ckflush clean the hollow-fiber membranes became attractive on the high fouling surface waters where microbes are of greatest concern. Then, the concern for integrity testing the elements grew. A test technique, based on gas diffusion rates and developed for pleated MF membrane cartridges used in pharmaceutical sterilization, was adapted. The hollow-fiber design lends itself readily to this test technique. The dual advantage of backflush cleaning and simplified integrity testing has made the hollow-fiber UF and MF configuration appealing.
Recent publications indicate that some currently perceive that the spiral design gives cause for extra concern for integrity. However, review of the literature, and discussion with manufacturers and users who have tested membrane over decades, does not support concern for actual integrity loss. More likely, the consensus in the industry is that hollow fibers are more likely to form leaks. However, where UF- and MF-class membranes provide adequate separation, the ability to easily apply the gas diffusion style integrity test gives them a perceived advantage for microbial removal applications.
Often UF- and MF-class membrane separation is not sufficient purification, so hollow-fiber UF/MF followed by spiral-wound NF/RO seems to be gaining favor as a treatment approach.4,5 Since NF and RO membranes are inherently capable of performing all the separations effected by the preceding UF and MF units, one naturally questions the need for both. Performing all separations with one step would clearly be a benefit, so good reason is required to add a second.
To achieve the ideal one-pass membrane treatment, either the development of lumen feed NF and RO membranes, or backflushable spirals should be most valuable. For the backflushable spirals, a foulant-resistant membrane (extremely hydrophilic perhaps) would make the job easier.
Capability of Spiral Elements and Systems for Microbe Removal
As stated, that the membrane itself is capable is not in question. The question is what can spiral elements, and the systems they are put into, reliably accomplish?
The answer is, they can remove all category of microbes and to a degree that will allow them to meet the SWTR (unaided; or in practice more likely as a key component of a series of treatment steps).
Controlled tests by independent microbiological laboratories show that a spiralwound ultrafilter can attain 5-log endotoxin reduction.6 Since endotoxins are only a portion of the cell wall of gram negative bacteria, they are more difficult to reject than bacteria itself. Another independent laboratory test demonstrated a 9-log reduction of bacteria (B.diminuta) in a similar spiral-wound ultrafilter.7
Of course, controlled laboratory tests can only demonstrate short-term performance, and do not account for the vagaries of a large commercial system. So how do spiral-wound systems perform for microbial removal in the field? A review of field systems, trade journals, and the research literature shows that they perform quite well.9,10
Relevant unpublished information includes the facts that since 1988, a New Jersey pharmaceutical manufacturer uses a standard design, large spiral-wound UF machine to make a 3-log reduction of endotoxins (to below the detection limit of 0.03 EU/m/L). The same machine produces 3 plus to 5-log reduction of total bacteria. This ultrafilter is part of a system for the production of USP Water for Injection, one of the most critical water quality applications. Osmonics® alone has over 30 RO systems installed for producing USP Purified Water, which has an 'action limit' to correct performance at a maximum bacterial count of 10 cfu's/mL. Also, Osmonics' internal testing of such systems shows that with only moderate attention to operational concerns, RO systems will regularly provide a 3-log reduction of bacteria. It is easy to extrapolate such performance to protozoan, cysts, and virus, which do not reproduce in water systems like bacteria do.
Crossflow membrane systems are well accepted in beverage plants, where stringent attention to water quality is also the norm. Single-pass system barriers have proven up to the challenge of microbial removal there as well. The Coca-Cola plant in Milwaukee received assurance, and proceeded accordingly, that because they had a spiral-wound NF system, they could bottle soft drinks using Milwaukee city water during the 1993 cryptosporidium outbreak.8
In U.S. municipalities, most of the brackish and surface water membrane installations use spiral-wound elements. Several studies have shown, through pilot testing on site at municipalities in California and Florida, that single-pass spiral-wound systems met the microbiological criteria requirements.5,9,10,11 Spiral-wound pilot systems have been recently installed, or there are plans for testing in Florida, Ohio, Virginia, Colorado, Minnesota; and undoubtedly others.
State regulatory agencies (and sometimes state laws) help implement the national EPA drinking water treatment regulations. Several states are reviewing field tests to allow spiral-wound membrane configurations for use in meeting the SWTR. The state of California regulations have been met in a pilot test of a spiral-wound NF membrane element. Effectiveness was demonstrated on canal water based on turbidity reduction, particle count reduction, a virus challenge, and with a cryptosporidium oocyst challenge where greater than 5-log reduction was obtained.12
There is considerable precedent and evidence that spirals are a suitable technology for use at municipalities. To these authors' knowledge's, no states have taken a position that spirals are not acceptable or not preferred.
The cost to operate a crossflow membrane system as large as municipalities install can be as important as capital. The simplest evaluation of cost from an energy view consists of the cost to run the pumps and the resulting permeate output. Each membrane configuration offers slight, but different advantages: Hollow-fiber elements have the least resistance for collecting the permeate. However, due to pressure drop limits caused by the lack of fiber strength, elements generally must be plumbed in parallel and not in series, increasing pumping volume required. Spirals have a fabric permeate carrier material, which presents some flow loss potential once the water has permeated the membrane (although minimal, it is rarely considered in energy cost). However, spirals easily withstand the pressure drops required to allow the concentrate stream to be used to feed another element in series. Generally, neither configuration is held to be inherently more 'energy-cost favorable' than the other.
The energy costs are in the required hydraulic flow and driving pressure, which relates to the membrane characteristics, both in its new and clean state and especially how fouling affects it. Real potential for energy savings is in reduced fouling. The ability to reduce both operating pressure and crossflow volume required depends on managing fouling. Of course fouling is based on the feedwater supply, but also the particular membrane's response to it (not all membranes foul equally). Both frictional pressure loss from flow and the energy effect of fouling are magnified when the operating pressure is lowered. With UF and especially MF, typical driving pressures are 50 psig to as low as 20 psig. Any pressure increase required to drive such a low pressure system becomes a significant cost factor. The element configuration, which mitigates fouling to the greatest degree, and most allows reversal of fouling (cleaning), gains advantage.
The feed flow in spirals is generally held to be turbulent at the membrane surface, whereas in fat hollow fibers the feed flow is laminar, allowing faster fouling. Hollow fibers can be backflushed, however, with the permeate to reverse fouling and a total throughput advantage may be attained.
A commercial system where spiral-wound elements can be backflushed in such a manner has not yet been developed. However, it has been done in the lab and is conceivable for commercialization. This is an area that should be explored more rigorously than it has been.
Further, should a membrane be developed which is so hydrophilic that the adsorption component of fouling is significantly reduced, then such backpulsing with spirals will be much easier to accomplish. At the same time, lower crossflow rates (i.e. higher recovery) and lower operating pressure may result. Such flat sheet form membranes (superphilic) are now being commercialized, although the two concepts have not yet been married in spiral-wound form. When such a concept is commercialized then lower pumping costs and even normal-flow operation (standard dead-end filtration, not crossflow) should follow.
Manufacturers have, of course, developed integrity tests since spirals have been around. The same is true for hollow-fiber elements. Early in the history of RO, only ionic species were considered, but removing organics and pathogens soon became important as the greater potential of RO was seen. An organic 'marker' test, where a certain passage level indicated poor quality membrane or an element flaw, was developed in various forms in the 1970's by manufacturers. Such tests, however, were never standardized across the industry, and each manufacturer kept them proprietary or at least unpublished. In 1980, the standards organization ASTM created a practice (standard number D3923) to explain this practice. As a 'practice,' it only addressed the concept in general and allows the user to set the specifics. Nevertheless, it describes two of the more common industrial techniques: an organic dye marker dynamic challenge, and a 'vacuum-hold' technique. The vacuum-hold test is based on the same principle as the bubble point, pressure hold, or pressure decay tests. (Originally developed to find flaws in sterilizing-grade pleated cartridges used for pharmaceuticals, and currently the basis of those tests being promoted for hollow fibers.)
Both these techniques are very viable and still in use today, although the industry has yet to agree to promoting a standardized approach.
That reluctance is changing with the increased desire to use membranes for large scale potable water treatment. Both the American Water Works Association Research Foundation (AWWARF) and the ASTM have begun initiatives in 1997. The AWWARF awarded a research contract to develop such integrity testing to a consulting engineering firm, which services municipal water treatment concerns. That research plan is based upon proving out integrity testing on spiral-wound elements.
Similarly, the ASTM committee D19 on water, subcommittee .08 on membranes and ion exchange, began an initiative in January 1997 to arrive at a formal 'method' standard, the most exacting and specific category of standards within ASTM. The task group plans to conduct round robin testing on these proposed methods to quantify and prove a correlation to leaks. This is a major undertaking by the ASTM, and they have canvassed regulatory agencies (including the EPA), the North American membrane manufacturers, and users such as municipalities for participants. This task group had an organizational meeting in June 1997, wherein representatives of the users groups and government agencies expressed their interest in seeing such a standard developed. Several manufacturers volunteered to share their background on such techniques. The current plan in ASTM is to have separate methods for different configurations if required, or a universal standard if feasible.
When the AWWA or ASTM creates such a standard for spirals, municipalities will probably use it. Until that time, individual municipalities and states are individually reviewing the qualification testing of spirals.
Use of membrane technology for treating potable water is growing in the United States and the rest of the world. There are several categories of contaminants, which require removal, and some classes of membrane can satisfy several of these (RO-NF). The choice of treatment technologies, including the choice of membrane configurations, will vary with the needs at each site. There are clear advantages, and disadvantages, for spiral-wound and hollow-fiber configurations. Energy cost (flow, pressure, foulin , and cleanability) and integrity testability are the two current marketplace drivers in the selection of the configuration. In response to these recently emerging market forces, new design approaches, new operational methods, new test methods, and even fundamental technology improvements are being developed. Such efforts are fairly young, and by no means complete. There are many reasons to believe that spiral-wound crossflow membrane elements will continue to be a major force in potable water production.
1. Taylor, 3.5., 'Drinking Water Regulation and Membrane Application,' Desalination and Water Reuse Quarterly: January - March 1995, Volume 4/4.
2. Paulson, D.J. et al, 'Crossflow Membrane Technology and its Applications,' pp 77-88, Food Technology - December 1984.
3. Paulson, D.J. et al, 'Design Innovation for Processing High Fouling Solutions with Spiral-Wound Membrane Elements,' Proceedings of North American Chemical Congress, Toronto, Canada, June 1988.
4. Vickers, J.C., et al, 'Bench Scale Evaluation of MF-NF Removal of Particles and Natural Organic Matter, Proceedings of AWWA Membrane Process Annual Conference, New Orleans, February 1997.
5. Lozier, J. et al, 'Meeting the Challenge of the New Drinking Water Regulations with Dual Membrane Treatment,' Proceedings of AWWA Membrane Process Annual Conference, New Orleans, February 1997.
6. Applied Microbiological Services Inc. Analytical Report of 7 May 87 on Endotoxin Challenge Testing of Spiratrex™ Spiral-Wound Ultrafilters (unpublished).
7. IER Inc. Analytical Report of 31 Oct 86 on Pseudomonas (now Brevundinmonas) Dimunata Challenge Testing of Osmo® UP2-PS Sepralators (unpublished).
8. Sfiligoij, E (Editor), 'Tales from the Cryptosporidium, Beverage Word, May 1994.
9. Thomson et al, City of San Diego Potable Reuse of Reclaimed Water: Final Results, Presented at the NWSIA 1992 Biennial Conference, August 1992.
10. Jolis, et al, Desalination of Municipal Wastewater for Horticultural Reuse: Process Description and Evaluation, Presented at the ADA (formerly NWSIA) l9~ Conference, September 1994.
11. Godman, R.R. et al, 'A Dual Water System for Cape Coral, J. AWWA, July 1997, Volume 89, Issue 7.
12. February 1995 Memo from the Senior Sanitary Engineer. Drinking Water Technical Operations Branch of the Department of Health Services.