Part 4: Membrane filtration
By Ed Butts, PE, CPI
The first three parts of this series on solids separation methods have covered strainer, barrier, centrifugal separators, cartridges, adsorption, oxidation, and precipitation methods of solids removal. We’ve discussed the most common filtration methods of removing particulate matter from water as well as many additional procedures that assist with the filtration process.
Now we’ll wrap things up by delving into one of the most popular types of solids separation methods currently practiced for process, potable, and industrial water supplies today— membrane filtration.
Introduction to Membrane Filtration
Most of the earlier applications of membrane filtration introduced in the 1960s used reverse osmosis for the specific need to desalinate seawater or brackish water in order to create drinking water.
Membrane filtration is now more commonly associated with and widely used for surface water supplies and treatment of potable and process water. Also, the many improvements seen in recent years in materials, technology, and types of membrane construction have increased the recent popularity in using these filtration devices in groundwater and other types of water and wastewater systems as well.
Membrane filtration was at one time often considered a “last gasp” method for the removal of salts and small and fine particulates and contaminants from drinking water—including bacteria and viruses where conventional filtration methods either failed or were inadequate. However, the improvements in the filtration method in recent years—to prevent premature failure caused by plugging, overloading, or the breakthrough of contaminants, plus the enhancement and knowledge of the need for proper pre-filtration and treatment—now makes the units more of a primary filter.
In reality, the recent improvements in the materials used, the improved design of membrane filtration cartridges, and using many units in parallel to lower individual flows makes the use of this system much less of a concern for plugging than in previous years. Membranes are now commonly used for the removal of various constituents including salinity, hardness and other dissolved inorganics, natural organic matter, synthetic organic contaminants, disinfection byproducts, and other common water contaminants.
In addition, selecting the proper membrane style and type for the specific application, along with implementing appropriate pre-treatment or filtration ahead of the membranes where needed, now allows the designer to tailor the application and design of a membrane system to a specific raw and finished water—and with proper maintenance, greatly extends the life and removal efficiency of membranes far beyond what would have been experienced during the earlier years of their use. Membrane systems for filtration and desalination, with the appropriate application and design, are now available in a range of flow rates from less than a gallon per minute for a single unit all the way up to 30 million gallons per day or more for systems with thousands of units operating in parallel.
When comparing membrane filtration to the various solids separation methods we have previously explored, it becomes fairly obvious membrane filtration is a form of the barrier method. What it does is place a physical obstruction directly in the path of the water and particulate and removes the solid through a sieving-type process.
Two main flow configurations are used as the sieves in membrane processes: cross-flow filtration and dead-end filtration.
In cross-flow filtration (Figure 1), comprising the majority of water and wastewater applications, the feed flow is introduced tangential to the surface of the membrane, retentate is removed from the same side farther downstream, and the rejected material is removed from the fluid and withheld in the membrane. Finally, the permeate flow is developed on the other side of the membrane. Depending on the specific design of the membrane, the feed fluid can be directed to flow from the center through the membrane, with the permeate discharged on the exterior side. This is referred to as inside-out flow.
Conversely, the feedwater can flow in from the exterior side of the membrane, through the membrane, and discharge into the center to flow outward in what is called outside-in flow.
In dead-end filtration the direction of the fluid flow is normal to the membrane surface. All membranes work under the same basic principle. The size of the sieve or port opening in a membrane is generally so small and tight that extreme values of applied pressure or head are required to force the water through the membrane. This is why most methods using membrane filtration for water treatment are typically operated using a pressure-driven function.
Typically, membrane filtration incorporates four different types of solids removal capability, here listed in the relative order of their potential of removing larger to smaller sizes of particulate matter or dissolved ions:
- Reverse osmosis
Microfiltration and ultrafiltration membranes are used extensively for filtration of process water and many industrial uses, as well as for pre-treatment of water, particularly ahead of reverse osmosis and nanofiltration units. Nanofiltration and reverse osmosis are more commonly used for water purification due to their superior ability to remove viruses and dissolved ions and solids, such as salts.
The relative size of particles removed for each type of membrane is shown on the chart in Figure 2 and are generally rated in micron removal capability. As you can see, the removal capability for bacteria and viruses with each method is indicated on the chart by the physical size of the contaminant in microns. The filtration ability of conventional particle filtration, including the majority of those discussed in this series, ends at about 1 micron in size, the generally accepted limitation to conventional filtration methods.
Microfiltration overlaps some of the range associated with conventional filtration, from a high particle size of 5 microns down to a particle size of just 0.1 micron. Ultrafiltration is further capable of removing material in size between 0.5 micron down to slightly less than 0.01 micron. Nanofiltration functions well in the removal of particle and dissolved ion sizes between 0.01 micron down to 0.001 micron. Reverse osmosis removes virtually all dissolved salts and ions between 0.001 micron down to 0.0001 micron and is the filtration process most associated with converting seawater to freshwater through removal of the saline constituents (ions) in the water.
Nanofiltration is actually a form of reverse osmosis with the distinguishing factor being the size of the retained ion. Since both processes use semipermeable (non-porous) membranes, there is no actual pore opening or nominal pore rating, although 1-2 nanometers (1 micron=1000 nanometers) for nanofiltration and less than 1 nanometer for reverse osmosis are often accepted as the typical removal efficiency. This also make both types much more susceptible to plugging and premature failure from particulate matter due to the extreme tight nature and lack of actual pore openings on the membrane. As such, the use of nanofiltration and reverse osmosis on a water supply with particulates is not to be used; the feedwater should receive an appropriate level of pre-treatment to remove these constituents before introduction into the membrane.
The type of particulate rejection for each class of membrane is shown in Figure 3. The straight, light blue arrows indicate the constituents generally not removed by and therefore pass through each membrane, while the right-angled, darker blue arrows indicate the constituents rejected or typically removed by each membrane.
In water treatment, the four types of membrane devices are generally classified as membrane filtration devices for the two processes involving microfiltration and ultrafiltration for larger particulate matter, and membrane desalination devices for the two processes with semipermeable membranes for removing dissolved salts and ions using nanofiltration and reverse osmosis. Due to their individual characteristics for their ability to retain very small particle sizes, the membrane classes of filtration devices can also remove various contaminants from water.
Although not usually recognized or certified for this specific use in potable water, each of the membrane types can nonetheless remove many of the most common types of contaminants from water supplies as well as lower the turbidity or cloudiness of the water. These typical contaminants again are indicated in Figure 3.
All four of the processes are capable of removing or lowering turbidity, most bacteria and cysts, and some viruses from freshwater supplies, including pathogenic and non-pathogenic types, although many of the smaller viruses remain too small for the larger filtration methods such as microfiltration and ultrafiltration.
This generally indicates the use of nanofiltration or reverse osmosis for effective virus removal. Since most protozoan cysts have a physical size between 2 microns up to 25-50 microns, modified methods of conventional particle filtration, bag filtration, and microfiltration are commonly used to remove these cysts, such as Giardia lamblia (typical cyst size = 4-7 microns) and Cryptosporidium (typical cyst size = 1-3 microns), from water supplies by using enhanced methods of water treatment or pre-treatment.
In addition, many bacteria, viruses, and cysts have elongated or irregularly configured body shapes common to a human torso. This means the method of particle removal must be capable of removing or retaining the smallest crosssectional size of the organism, not simply the average size, which in some cases could easily pass through a larger pore size in a membrane and into a water supply.
This explanation relates to the earlier discussion we had in Part 1 regarding pore size and the removal rating of filters. The pore size correctly refers to the actual opening size of the pores (holes or openings) in a filter. This may be reported as a minimum pore size (smallest measurable hole), maximum pore size (largest measurable hole, the most meaningful classification for membranes), or a pore size range (5-10 microns). In many cases the actual pore size rating of a given filter may be much larger than the removal rating of a given filter.
The removal rating refers to the statistical probability of a filter’s ability to remove a certain size particle when challenged under controlled conditions. This should not be confused with the actual pore size of a filter. There are two types of filter and membrane ratings: nominal and absolute.
These terms are misused to a great extent in filter claims and marketing literature, which can mislead the designer or user. A nominal rating is attached to membranes that can be shown under controlled conditions to remove an acceptable statistical amount of particles of a certain size, even though the actual pores or openings of the membrane may be much larger than the particles being removed.
The typical way a nominal rating is reported would be: “Removes >99.9% of particles 3μm (microns) or larger.” This means under field conditions a user can be confident the membrane will remove greater than or equal to 99.9 percent of pathogenic organisms larger than 3 microns. This information is obtained by challenging filters with test waters containing suspensions of 3-micron spheres. A similar claim for cryptosporidium or giardia would require a challenge using live cryptosporidium or giardia organisms.
An absolute rating can only be applied to a filter or membrane in which the end user can actually determine the physical size of the largest pore (opening). Such a membrane can be integrity tested by using a nondestructive test method and the resulting data used to determine the actual size of the largest pore.
Either nominal or absolute ratings can be applied to membrane filters due to an industry requirement of using a definable pore size in membranes. This type of certification provides the designer with a confidence not usually associated with a strainer type of filter or even most barrier types.
Membranes are constructed using one of three basic configurations: spiral wound (Figure 4), hollow fiber (Figure 5), or tubular (Figure 6). Each type has specific advantages and disadvantages that must be evaluated for the feedwater, application, and desired quality of the finished water.
The two styles most commonly used in water treatment are the spiral wound and hollow fiber. Microfiltration/ultrafiltration (MF/UF) modules are typically arranged in a hollow fiber configuration in which hundreds to thousands of elongated and straw-shaped membranes are bundled together. MF/UF and nanofiltration/reverse osmosis (NF/RO) membrane units are generally placed in skid-mounted rack assemblies with multiple modules (MF/UF) or within pressure vessels (NF/RO) that contain the membranes.
MF/UF membrane modules can also be configured as either encased or submerged systems. Encased membranes operate under a positive pressure and can use either inside-out or outside-in flow patterns. Submerged membranes are operated under negative pressure using filtrate pumps to create a suction in an inside-out flow pattern. NF/RO membranes use a spiral wound configuration, consisting of a series of flat sheets of membrane material rolled in a cylindrical shape, called an element. As water permeates through the membrane, it is collected in a center tube. The water that doesn’t permeate contains the rejected contaminants at higher levels, becoming the concentrate stream.
Typically, the hollow fiber membrane operates using a lower pressure than the spiral wound, between 5-50 psi. The spiral wound style, however, requires much higher pressure for the feedwater to permeate the membrane—values between 75 up to 1400 psi are usually applied. High pressure, multistage booster pumps are often used to generate this extreme pressure. Both types also require some form of pre-filtration. For the hollow fiber configuration an inline strainer type will generally suffice, while the spiral wound type requires a prefilter with 1- to 5-micron removal capability.
Finally, it is critical all aspects of a membrane’s tolerance to chemicals—such as chlorine or other acid or scale inhibitors used to clean the membrane—be fully evaluated before selection is made. Many membrane materials dissolve rapidly when exposed to even low concentrations of chemicals, and this factor must always be examined in advance.
In the final analysis, selection of a specific membrane type can only be made following a technical review of the feedwater (source), the desired finished quality, allowable rejection percentage, chemical tolerance, waste disposal methods, pressure configuration, flow rate, and—of course, cost.
Membrane Operation and Performance
In addition to monitoring water quality, operators must continually gauge the membrane’s performance. Membrane systems in a water system are often tied into a supervisory control and data acquisition (SCADA) system just like a conventional filtration system would be, but instead of monitoring run times and filter head loss, operators monitor the pressure on both sides of the membranes.
Membrane performance usually changes over time, generally with a decline of flow, because of fouling or plugging. For MF/UF membranes, fouling caused by filtered particulate matter is typically reversible and removed by backwashing procedures. However, dissolved organics and biological agents produced by microorganisms may cause severe fouling that necessitates chemical cleaning.
In addition to organic and biological fouling, NF/RO membranes are also susceptible to scaling deposits caused by the precipitation of silica and other soluble salts. Also, as previously noted, NF/RO systems aren’t specifically designed to remove particulate matter, which will rapidly plug the membranes. In these instances, adequate methods of pre-treatment must be used to catch the larger offending particulate before the membrane.
When gauging membrane performance, operators must routinely monitor the three operating elements: the flux (= flow per unit size of membrane area), pressure, and temperature. For MF/UF systems, the transmembrane pressure can increase up to a product-specific threshold determined by the manufacturer, at which point the membranes must be cleaned.
Wastewater and Residuals Handling
A consequence of intermittent or continuous membrane use is the creation of byproducts—or a residuals stream— with high concentrations of contaminants removed by the membranes. For MF/UF processes, the primary residuals stream created is the periodic backwash flow; for NF/RO it is a continuous concentrate stream of salts. Both types of systems include periodic chemical cleaning wastes as well.
Recovery is the percentage of feed flow converted to treated water—the rates varying depending on the feedwater quality. Recoveries for MF/UF are relatively consistent and predictable with rates almost always exceeding 90% and typically in the 95% range.
NF/RO recoveries exhibit much greater variability, ranging from a low of 35% (for seawater with high TDS) to as high as 90% (for brackish groundwater applications with low TDS and limited concentrations of scaling ions). The recovery percentage of NF/RO systems can often be increased by using a solution of pre-treatment acid or a scale inhibitor to control scaling of the membrane.
MF/UF residuals are generally similar to those from conventional surface water treatment plants and are often discharged to a nearby sanitary sewer where a service with adequate capacity is available. Disposal of the highly saline NF/RO concentrate is typically much more of a problem. For those cases in which the concentrate doesn’t represent a significant portion of the local or regional wastewater flow, discharge to a sanitary sewer may be possible, as the blend from domestic sewage will generally dilute the impact from the salinity. Discharge of the concentrate to a septic tank or drain field is typically not recommended because the heavy loading of salts can rapidly destroy the biological action within the tank and cause plugging of the leach lines and percolation zone.
An alternative to conventional membrane filtration for water is bag filtration. This process, a version of the barrier method, generally uses a skid-mounted assembly consisting of two to three cartridge type filters (bags) with progressively greater filtration capability in series to remove pathogenic disease-causing organisms, such as bacteria and protozoan cysts including giardia and cryptosporidium.
Each individual unit is capable of a flow rate of 10 to 20 gpm depending on the application, the bag’s surface area, filtration log rating, and manufacturer and regulatory agency’s specific design limitations.
Additional units are assembled using multiple units in series to create trains to improve the filtrate quality and in parallel operation to increase the overall flow rate. This relatively new introduction to filtration systems represents an alternative and low-cost solution to conventional and membrane filtration systems for many potable water systems, particularly for small to medium water systems with typical flow rates between 10-200 gpm.
Bag filters are manufactured by various firms throughout the United States and many are now certified by the EPA and several state agencies for up to 2-3 log (99%-99.9%) removal of various contaminants. The system typically operates with a pre-filter, which is used to remove larger material feeding into a post-filter designed to remove contaminants down to 2-3 microns. The degree of pre-filtration is normally determined based on the raw water quality and what is needed to extend the service life of the final filter since that the filter is usually the most expensive.
The use of bag filters are generally more predictable than membrane systems since the filtration area is well defined and the raw and finished water qualities are usually known. The system is quite versatile in practice and application and they can be found for potable water, process, and industrial water filtration for pressurized and open systems. In pressure systems, the required pressure for the final filter varies, but is usually around 2-3 psi in a clean state during operation up to a terminal pressure between 15-30 psi. When considering using bag filtration, it is imperative the designer doesn’t try to match one brand of an approved filter bag with an unapproved canister. It will usually not work. The procedure for gaining NSF or UL certification typically requires the filter be challenged as an assembled unit. This is the only reliable way to be sure the filter system will effectively provide the log removal of the contaminant stated by the manufacturer.
So What Should We Use?
As we close out this series, there is one more question to be asked: “So what do we need to use for our system?” This is probably the single most asked question in filtering applications— not which mesh or micron size, but which actual kind of filter should be used. Although numerous criteria can and should be applied when considering individual types of filtering units, let’s look at the three most important considerations.
1. What are we taking out?
Obviously, this is a primary consideration as this question often controls the device itself. For example, if removal of Giardia lamblia is the goal, then a filter capable of removing particles with the cyst’s physical dimensions, or around 4-7 microns, should be the design criteria that controls the pore size or membrane capability. If multiple organisms or particulate must be removed, the designer then must also incorporate the added burden from these particles or use an effective pretreatment process.
2. What is the needed removal rate?
It is simply not enough to state that giardia must be removed from a water supply. Regulatory agencies, specifically the EPA, has created and enforces a minimum level of the removal or deactivation of these cysts. Many filters may have the ability to remove 90% of giardia, but the normal requirement is removal of 99.9% of the cysts. Due to the varying shape and concentration usually found in water, this level often requires either chemical addition to enable flocculation or extremely tight filtration to accomplish this. Regardless of what kind of organism needs to be removed, you must also know how many of the organisms must be taken out.
3. What is the flow rate?
The design flow rate controls many elements of a filtration device, namely the inlet and outlet sizes, net screening or membrane filtration area, and whether the device is an inline pressurized unit for lower flow rates, a suction screen for a 1000 gpm irrigation pump, or a large gravity type of unit for larger flows.
All three considerations and several more comprise the majority of the elements involved in selecting a filtration method. To do the job right, all these issues and more must be determined. We have just scratched the surface in the subject of solid separation processes. The primary element to take away from this discussion is no type of filtration should be taken lightly or performed simply by using off-theshelf components without a full examination of the water supply, the type and degree of the rejected materials, waste disposal, cost, and operational considerations. An extended pilot study is often needed to be certain about all the above considerations.
I hope you found the information in this series of interest and can apply some of it to your next filtration project.
Until next month, work safe and smart.
Ed Butts, PE, CPI, is the chief engineer at 4B Engineering & Consulting, Salem, Oregon. He has more than 35 years experience in the water well business, specializing in engineering and business management. He can be reached at firstname.lastname@example.org.