April 3, 2004

Reverse osmosis (RO) is a desalination technique that is fast gaining attention due to its simplicity in operation and significant improvements in membrane technology.

Despite breakthroughs in membrane technology, the performance reliability of the RO desalting technique has been greatly affected by membrane fouling. With that, replacement rate of the RO membrane and downtime of the plant increases. As a result, operating cost also inflates.

Membrane fouling is attributed largely to the seawater and RO process chemistry. To control and reduce the extent of membrane fouling, much can be done in the area of pretreatment upstream of the RO membrane in providing better quality feed water.

With proper feed pretreatment, fouling potential of the seawater feed can be reduced, leading to better survivability of the RO membranes.

However, performance of media filtration, which has been the conventional pretreatment technique for many years, is still lacking in many aspects. Intensive consumption of chemicals and inconsistency in performance are some of the main disadvantages of media-filtration.

Recently, a double membrane barrier concept has emerged. By installing a microfiltration (MF)/ ultrafiltration (UF) membrane upstream of the RO membrane in an integrated membrane system (IMS) to filter out bigger suspended solids, the feed water quality can be improved. As a result, the useful life of the RO membrane may be extended. There are even cases where performance of RO systems exceeded its initial design specifications after adopting the membrane pretreatment techniques.

In case of Pakistan, research into what would be the right choice in desalinzation should focus on the evaluation of performance of various membrane filtration techniques in the pretreatment of seawater.

In order to ensure results close to actual operating conditions, coastal pilot plant should be conducted. The experimental findings would thus help chalk out a comparative analysis of the different types of membrane pretreatment systems.

For the membrane pretreatment techniques, flux and pressure trends should also be considered, as they have important implications on the stability of performance. Silt Density Index (SDI) must be used as a primary yardstick of filtrate quality during the pilot testing of the pretreatment systems.

Conventional system
Besides fluctuations in SDI of the filtrate produced by the conventional pretreatment system, other major concerns include extensive chemical consumption and frequent replacements of micron cartridge filters. In the case of chemical additions, optimal dosing rates have to be tested for each site location.

Any changes in dosing rates required are often reliant on the experience of the plant engineers, making automation of the pretreatment process difficult. Undesirable chemical reactions may also occur between the polymeric coagulant used and the organics present in seawater. This may lead to fouling and degradation of the RO membranes. Therefore, looking for an alternative pretreatment techniques which can remove colloidal particles and consistently produce filtrate of low SDI is a natural extension from the current research on conventional pretreatment methods.

Alternative methods
The implementation of UF and MF membrane pretreatment techniques is considered below.

Microfiltration: Microfiltration (Mf) can be defined as the separation of particles of one size from particles of another size in the range of approximately 0.01 µm through 20 µm. The fluid may be either a liquid or a gas. This technology for solids separation is called “micro” because it uses membranes with pore size ranging from 0.04 to 20 microns

Ultrafiltration: This is a process in which a porous membrane is used to separate or reject bacteria, viruses, large organic molecules, colloidal and particulate matter. Similar to other membrane processes, ultrafiltration is a pressure-driven process. High permeability of the ultrafiltration membrane and negligible osmotic effects allow the UF process to operate at relatively low pressures

Typically, UF modules incorporate capillary or hollow fibre as the membrane structure. The benefit of this type of construction is that it allows for backwashing of the membrane when the filtrate or product flow rate has decreased due to accumulation of material on the membrane.

The UF process can be operated “dead end” or direct flow, which is 100 per cent conversion of the feed fluid to filtrate, and cross flow. Cross flow eliminates zero flow areas without increasing the concentration of the feed stream and minimizes the boundary layer. Depending on manufacturer, configuration, material, pore size, dimensions and operating conditions vary.

Encased membrane systems: It is also known as Pressure Membrane System due to positive pressure used for the operation of this system. The various manufacturers are standardizing dimensions to allow for increased flexibility within a particular system. This allows for relatively simple changes to a system due to alterations in feed water quality, system performance standards, etc.

Some manufacturers provide their modules with end structures that are already configured to accept the required piping connections, thus eliminating the need for housing. This can save a significant amount of capital cost.

An encased system may utilize a cross-flow design, which allows for a high flux with minimal solids build-up on the membrane surface, or it can operate in “dead-end” mode where all water processed through the filter comes out as filtrate.

Submerged membrane systems: In these systems, the membrane fibres are immersed in an open tank and are exposed to the feed water. Groups of membrane fibres are bundled together in racks or modules. The membrane fibres are slightly longer than the distance between the upper and lower attachment points, allowing the membrane fibre to shake during operation to dislodge accumulated solids. Multiple modules can be submerged in the process tank, depending on the permeate flow required.

Submerged membrane systems can effectively replace clarifiers and multi-media type filters found in conventional water treatment plants and are capable of operating effectively and continuously in high-solids environments.

In addition, these systems can be particularly cost effective if an existing tank/basin can be utilized.

Construction material: Membranes constructed of a wide variety of materials are available. These materials vary widely in their chemical and mechanical properties including mechanical strength, burst pressure, oxidant tolerance, VOC tolerance, pH operating range, and so forth.

The end user must be aware of the strengths and limitation of each material type and ensure the selected material is compatible with raw water quality, pretreatment requirements, and other operating conditions.

Some of the commonly available materials include polypropylene (PP), polyethersulfone (PES) polyvinylpyrrolidone (PVP) blends, polysulfone, polyvinyideneflouride (PVDF), cellulosic derivatives (CD), and polyacrylnitrile (PAN).

Pretreatment: It can be employed prior to membrane treatment to control fouling, to provide additional treatment, to meet manufacturer warranty requirements, and to satisfy regulatory requirements. Types of pretreatment employed depend on the nature of the raw water quality and the goal of the pretreatment.

Pretreatment can reduce the solids loading in the membrane feed stream, permitting higher flux rates and in turn reducing initial capital costs by reducing the required membrane area. A reduced solids loading also has the effect of reducing membrane operating costs because of a reduced trans-membrane pressure requirement.

Cross flow vs. dead-end filtration: The direction of feed water flow, in relation to the membrane surface, determines the type of filtration in a membrane system. In a dead-end filtration system the feed water effectively contacts the membrane surface at a perpendicular angle. In contrast, feed water in a cross-flow system flows parallel to the membrane surface. Dead-end flow can also be achieved in a cross-flow system by simply closing the discharge valve. Feed water flow direction accounts for a variety of differences between the two systems — particularly, the two modes of operation may experience; differences in fouling rate, flux and recovery, and finished water quality.

In cross-flow mode, solids are continuously flushed from the system resulting in less frequent back-pulses and backwashes, and possibly longer membrane life. In dead end operational mode, solids build up in the system, thus more frequent backwashes and back-pulses will be required. Some systems are capable of operating in cross-flow or dead-end mode.

Inside-out vs. outside-in flow: Inside-out flow and outside-in flow refers to the direction of feed water passing through the membrane, as well as the orientation of the feed water in relation to the membrane surface. For instance, in an outside-in system, the feed water surrounds the membrane, and the filtrate is collected from inside of the hollow tube fibres (lumen). The outside-in scheme has the advantage of a larger membrane surface area, which allows for a slightly higher flow than the inside-out model while still maintaining the same flux rate and solids concentration.

An inside-out system places the feed water inside the fibres, and the filtrate is collected on the outside of the membranes. The feed water enters the fibres at one end of the membrane element, and the discharge passes through the element and exits the fibres on the other side. The filtrate is collected inside the element on the outside of the fibres.

Both methods have been shown to operate successfully by different manufacturers. What the user should consider in evaluating the best options for a particular situation are the backwashing implications of the two different configurations. The backwash cycle for virtually all systems consists of introducing clean filtrate back into the modules through the filtrate conduit within the module. Thus the inside-out modules backwash from outside-in and vice versa. Depending on the water quality, the configuration of the modules and the water treatment goal, one type of configurations and backwash scheme may be more suited than another. It may only be determined by pilot studying which may be the most effective for a particular application.

Cleaning cycles
Cleaning MF and OF systems is very similar in nature to RO cleanings. However, the MF/UF systems have the added advantage of backwash and back-pulse during the cleaning cycle. The MF/UF membranes, depending on the material of construction can be cleaned with caustic solutions to remove organics. Citric acid or other acid solutions are used to remove scalants, and oxidants may be used to remove biological or inorganic foulants, if the membrane material is tolerant of oxidants. The chemicals to be used depend on the type of fouling and the material of membrane construction. Cleaning cycles can be automated and implemented into the system design. A typical cleaning cycle can last for anywhere from 10 minutes to 2 to 3 hours per bank, including soaking and rinsing times. Cleaning frequency for surface water applications can again vary tremendously, from once per day to about 4 to 6 weeks. Cleaning procedures, durations, and frequencies will vary significantly depending on feed water quality, system design and operation, flux rates, and recovery rates. Cleanings may be automated and performed in place or they may be manual with the modules actually being removed from operation and cleaned in a special cleaning area. Submerged systems have the ability to remove cartridges service and clean them in a special tank. Encased modules are usually cleaned in place, but can be removed to a special cleaning rack.

Backwash procedures
All UF/MF systems incorporate a backwash cycle, whether operating in dead-end or cross-flow modes. A backwash cycle commonly operates on a preset timeframe, usually about every 15 to 60 minutes and is initiated by the PLC controlling the system. A typical backwash lasts around one to two minutes. The flow rate can range from lower than the feed flow to as much as ten times as high as the feed flow. All elements in a given array are backwashed together. The backwash water is typically collected in a backwash tank. The tank ensures the backwash fluid is released to atmospheric pressure. Backwashes can also be initiated when the trans-membrane pressure reaches a predetermined limit as recommended by the system manufacturer.

Typically, filtrate water is used for the backwash. The filtrate flows out of a tank and through the filtrate end of the membrane, then to discharge. The backwash releases solids that are attached to the feed side of the membrane surface. Backwashing improves the operational flux of the system to maintain desired performance. It is recommend to include a forward flush to drain to ensure that solids dislodged during backwashing are thoroughly flushed out of the system prior to directing filtrate to the filtrate tank.

Experience has shown that this will greatly reduce turbidity spikes that may be seen after backwash cycles. Compressed air, as in the Memcor (US Filter) system, can also be used for backwash purposes. The compressed air, usually less than 100 psi, is introduced on the filtrate side to remove accumulated suspended solids. Many systems also have the ability to have a continuous gas stream flowing through the membrane. This provides added protection against solids build-up.

The membrane systems that employ outside-in flow during normal operation generally use a back-pulse feature to dislodge solids from the outside surface of the membrane fibres. During a back-pulse, membrane permeate is drawn from a holding tank and is pumped in reverse direction, from the inside of the fibre out effectively dislodging solids from the outside of the membrane. The back-pulse water can be chemically conditioned with acid, caustic, or biocide to further clean the membrane fibre. Back-pulses typically are initiated automatically by the PLC controlling the system and occur at 15-30 minute intervals with a duration of 10-30 seconds.

Performance characteristics

Flux: Contrary to reverse osmosis, the flux in a MF/UF membrane system is not a characteristic of a specific manufacturer so much as it is limited by the following parameters:

— Raw water quality (temperature, solids content, etc.)

— Efficiency or existence of the pretreatment process.

— Maximum trans-membrane pressure as dictated by the physical properties of the membrane polymer and limitations of manufacturer’s equipment.

— Acceptability of the resulting cleaning frequency

All of these factors are interrelated and will therefore dictate the flux rate achievable on a case by case basis. Flux rates though can be as low as 5 GFD (Gallons per day per square foot of membrane area) for challenging waters, and flux rates have been documented over 90 GFD for very clean feed waters. Average flux rates range from 25-40 GFD.

Recovery: Similar to flux, recovery rates in a MF/UF membrane system are not a characteristic of a specific manufacturer so much as recovery is limited by the following parameters:

— Raw water quality (temperature, solids content, etc.).

— Efficiency or existence of the pretreatment process.

— Maximum trans-membrane pressure as dictated by the physical properties of the membrane polymer and limitations of manufacturer’s equipment.

— Acceptability of the resulting cleaning frequency

High recovery rates generally require high trans-membrane pressures to operate higher cleaning frequencies because the membranes are forced to run in a higher solids environment. All of the above factors are interrelated and will therefore dictate the recovery rate, achievable on a case by case basis. Recovery rates can be as low as 50 per cent for challenging waters, while recovery rates have been documented at up to 99% for very clean feed waters. Average recovery rates range from 92% to 95%.

Particle rejection: The main difference between membrane filtration and conventional filtration is that membranes reject particles based on size exclusion as opposed to media depth filtration. For all practical purposes, MF/UF membranes will produce a constant permeate water quality regardless of feed water conditions because they act as an absolute barrier. This can be expected since most suspended solids present in raw water streams are larger than the pore size of most MF/UF elements. Permeate turbidity can be expected to be less than 0.1 NTU or lower and SDI less than 0.4. However, breaches in integrity such as a broken fibre or ruptured seal, or imperfections in the membrane due to the manufacturing process can allow particles to enter the permeate stream.

While it has been said that MF/UF membranes pose an absolute Carrier to suspended solids, it is more appropriate to express solids removal in terms of log removal. Most membranes are capable of up to 6-log pathogen (protozoan cysts and bacteria) removal, though this will vary depending on the membrane manufacturer. Virus rejection varies significantly between manufacturers with some systems removing virtually no appreciable amount of viruses while others report greater than 6-log removal.

Factors
The factors affecting fouling and cleaning are:

Frequency — Fouling is the most serious disadvantage of pressure-driven membrane separation processes. Membrane fouling results in a decrease in flux and an increase in energy consumption and feed pressure. Fouling will occur in any MF/UF system, regardless of the membrane polymer, system manufacturer, and mode of operation. Fortunately, fouling can be effectively controlled through the proper use of pretreatment processes, chemical additions, and proper system design and operation. The following factors are the most common causes of membrane fouling and can be controlled by proper design and operation.

Equipment malfunctions - Malfunctioning equipment can increase cleaning frequencies. For example, malfunctioning level control equipment in a submerged membrane system can allow membranes to operate in partially full tanks, thus accelerating membrane fouling.

Pretreatment chemicals: Pretreatment chemicals must be evaluated for their compatibility with the membrane materials.

Temperature: The effect of temperature on permeates flux is well documented in the membrane industry. Decreases in temperature decrease flux, increase TMP, and increase fouling rates. Cleaning frequencies can double with a 15-degree C change in water temperature.

Flux: Increasing flux increases cleaning frequency.

Recovery: Increasing recovery increases cleaning frequency due to higher solids concentrations in the feed water. For example, the solids concentration factor is 12.5 at 92% recovery and is 100 at 99%recovery.

Cake formation: Accumulation of particles near the surface of the membrane forms a cake layer which can be considered a second membrane. The cake layer can enhance the particle rejection characteristics of the membrane by depth filtration and adsorption. However, the cake layer increases the resistance to flux and leads to higher trans-membrane pressures. This type of fouling is reversible and can be removed by cleaning and back-pulsing.

Adsorption: Organic matter present in the feed water can lead to .membrane fouling either be adsorbing into the cake formation or by adsorbing into the membrane itself. Some membrane polymers exhibit higher adsorptive rates than others, depending on the hydrophilicity of the polymer. Pretreatment can be optimized to remove as much organic matter from the feed to extend membrane life, and chemical cleaning can be optimized to remove adsorbed foulants.

Metals precipitation: Though metals precipitation in MF/UF membranes is not as prevalent as in NF/RO processes, fouling due to metals precipitation has been documented. Calcium carbonate scale can form on the permeate side of membranes in scaling waters where carbon dioxide is driven off as part of the process. In operations using membrane filtration for iron and manganese removal, precipitation of these metals on the concentrate and permeate side can occur. Post-precipitation of iron and aluminum associated with metallic-based coagulants has been observed and can lead to fouling on the permeate side and increased particle counts in the permeate.

Costs
Capital costs of membrane system vary depending on the manufacturer, the scope of supply, the design capacity of the system, and the level of competition during bidding. The capital costs discussed in this section include the membranes, skids, racks, compressors, blowers, pumps, piping, instrumentation, controls, and other components necessary for a complete and operable system. The capital costs are very competitive for encased systems up to 10MGD, and submerged systems are more cost-effective for systems larger than 10MGD.

Membrane treatment is one of the most rapidly advancing water treatment technologies. MF/UF membrane processes have gained wide acceptance in the drinking water industry because of their ability to produce a high-quality and consistent product water. More recently MF/UF membrane filtration has gained acceptance as a pretreatment for nanofiltration (NF) and reverse osmosis (RO) for seawater applications. The designer or end user is overwhelmed by the variety of technologies that exist. There are advantages and disadvantages for the many various technologies that are available in MF/UF systems. With proper study and review of data versus available membrane capabilities, a designer will be able to design a system for the end user that will be tailored to the application and will operate efficiently and cost-effectively for years.

The writer is director of the International Desalination Association (IDA)

Please Visit our Sponsor (Ads open in separate window)