Advances in Membrane Performance

Transcription

1 Advances in Membrane Performance Introduction Rich Franks, Mark Wilf and Craig Bartels Hydranautics Since manufacturing the first RO elements in the 1960s, steady advances in RO technology have significantly reduced the cost of desalinating water. One major leap forward came in the late 1970 s with the introduction of the composite polyamide chemistry which improved both salt rejection and membrane permeability compared to that of the widely used cellulose acetate membrane. Other advances came in the 1990s with the introduction of the energy saving polyamide (ESPA) membrane and low fouling composite (LFC) membrane. But the new millennium has seen a maturation of membrane technology. During the past five years, improvements in RO performance have come from less dramatic, more incremental advances in both the design of the RO element and the formulation of membrane chemistry. Most of these advances in membrane performance come from enhancements in existing membrane chemistries and from optimizing the design and construction of the widely used spiral wound element. Though these recent advances do not equal the impressive breakthroughs of the past, the incremental improvements of today continue to generate significant benefits for the end user, especially when factored over the life of the RO plant. Improving the RO Element Recent improvements in the design and construction of the RO element have come by focusing attention on the details of glue line placement, feed spacer size and configuration, and the permeate carrier design. By optimizing these components, membrane area is increased and pressure losses are reduced, leading to an improvement in element productivity apart from any modification to the membrane chemistry. The move by most RO element manufacturers to automate element production has also led to an increase in element performance by precise control of fabrication variables. Automated production contributes, among other things, to a 10% increase in the active membrane area of a single element. This in turn leads to higher productivity. Similarly, greater membrane area, and therefore greater productivity, is achieved through a modification of the element s anti-telescoping device (ATD). The old style, Figure 1, left an empty volume between each element installed into a pressure vessel. By switching to a flush cut design, Figure 2, 3% more active membrane area was added to each element. Another factor leading to higher productivity involves the tradeoff between the number of membrane leaves wound into an element and the length of each leaf. As leaf length increases, the number of leaves must decrease. Recent improvements in element design have optimized the balance between leaf length and leaf number for the current generation of higher flow membranes by including more leaves of shorter lengths. The design of the permeate carrier has also been optimized to accommodate higher productivity. A tradeoff exists between permeate channel width and pressure losses incurred from water flow through the permeate channel. The expectation was that wider permeate channels would create a greater cross sectional area and result in lower pressure drops as the permeate flows to the core tube. However, investigations have shown that the wider permeate channel leads to embossing of the membrane into the channel resulting in a reduced cross sectional area and greater pressure losses. The embossing also fatigues the membrane and leads to shorter life.

2 Old Seal Carrier/ATD with Internal Interconnector New Seal Carrier/ATD with Internal Interconnector Figure 1 Figure 2 Comparing the performance of the newer ESPA2+ with that of the conventional ESPA2 illustrates some of these improvements in element design and manufacturing. The ESPA2+ benefits from the latest technological advances so that it can produce 33% more permeate than the ESPA2 at comparable conditions. Like ESPA2, the ESPA2+ can be used on high or low salinity brackish streams when the highest rejection is required. The higher productivity ESPA2+ translates into either capital savings or operational savings depending on the specific design requirement. For example, if an RO system is designed using the same number of ESPA2+ elements as ESPA2 elements, the increased area per element will lead to lower pressures and a lower rate of fouling which in turn leads to lower operating cost. If, however, capital cost savings is preferable, fewer ESPA2+ elements can be used in an RO system to achieve the same flux as a system designed with more ESPA2 s. Fewer elements, less pressure vessels, and less piping translate into additional capital savings. Lower Differential Pressure (LD) In addition to optimizing glue line placement, permeate carriers channel width, and ATDs, the brine spacer has also been modified to decrease pressure losses while still providing sufficient turbulence to control concentration polarization. The advantages of employing a thicker brine spacer of 31 mil over the 26 mil include reduced pressure losses and improved cleanability. But going beyond merely increasing brine spacer thickness, the new LD element series optimizes the construction and geometry of the brine spacer to further decrease pressure losses. To illustrate this difference, Figure 3 compares the pressure loss of the enhanced 34 mil spacer LFC3-LD with elements using standard 26 mil and 31 mil brine spacers. Due to the other improvements in element design and automated manufacturing as discussed above, the increase in brine spacer thickness is achieved without sacrificing surface area. The introduction of the LD to the low fouling membranes series gives the RO designer more options to address a range of fouling issues associated with challenging surface and waste waters. Depending on the fouling characteristics of the water being targeted and the desired pressure and rejection, the RO system designer can choose from the low fouling membranes listed in Table 1 to obtain the following advantages: Combined hydrophilic membrane chemistry and neutral surface charge resulting in lower affinity to organic fouling. Low feed pressure, low differential pressure, and/or high rejection. A significant increase in membrane life and system operation. Prolonged periods between cleanings and a significant reduction in cleaning costs. An increase in membrane life and system operation.

3 Differential Pressure(psi) LFC3 26 mil 31 mil LFC3-LD 31 mil Brine (gal/min) Figure 3: Differential pressure across a single 8 inch element with brine spacer thicknesses of 26 mil versus standard 31 mil versus LFC3-LD Table 1. Comparison of Low Fouling Composite Membranes. Product Brine Application Type Spacer LFC1 LFC3 LFC3-LD 26 mil 26 mil 31 mil Municipal and industrial surface and waste water applications where low pressure is a priority Municipal and industrial surface and waste water applications where high rejection is required Municipal and industrial surface and waste water applications where high rejection and low differential pressure is required GPD (m 3 /d) Performance Rejection (avg.) 11,000 (41.6) 99.5% 9,500 (36) 99.7% 11,000 (41.6) 99.7% Large Diameter Elements The industry standard for the spiral wound elements is 8 inch diameter by 40 inches in length. This standard has been in place for over 20 years. However, a study released by a consortium of membrane manufacturers and consultants documented the potential economic advantages for brackish, groundwater, and seawater desalination by going to a 16 inch diameter standard for the next generation of spiral wound elements (Bureau of Reclamation, 2004). The study concluded that construction cost savings of 18% to 27% could be realized by switching from the 8 inch to the 16 inch. Since the release of that study in 2004, most membrane

4 manufacturers have unveiled their versions of the large diameter elements. The Hydranautics 16 inch comes in both a seawater version (SWC3 16x40) and an energy saving brackish version (ESPA2 16x40). Table 2 below compares the 16 inch element performance with its associated 8 inch version. The 16 elements have been or are currently being field tested in large diameter RO pilot systems. Table 2. Comparison of 16inch and 8inch spiral wound element performance at standard test conditions (800 psi and 32,000 ppm NaCl for SWC. 150 psi and 1500 ppm NaCl for ESPA) Product Area (sqft) Area (m 2 ) (gpd) (m 3 /d) Rej (%) Dry Weight (kg) SWC , SWC3 16x , ESPA , ESPA2 16x , RO Membrane Improvement Coupled with the improvements in element design and manufacturing, advancements in the existing polyamide membrane chemistry have yielded state of the art performance. These improvements result in higher permeability and therefore lower pressures. Modifications to the membrane chemistry also lead to customized membrane targeting a specific application or the removal of a specific constituent in the feed stream. Seawater (SWC) As an example of these improvements in membrane chemistry, Table 3 below illustrates the evolution of the seawater (SWC) membrane over the past four years. The increase in element permeate flow in going from the SWC3 to the SWC3+ came from design and manufacturing advances discussed above, but a significant portion of the increase in permeate flow as seen in the SWC5 came from further enhancing membrane chemistry. Table 3. Evolution of the seawater RO element. Performance is based on testing at standard test conditions of 3.2% NaCl, 800 psi, 10% rec, and temperature of 25 C. Product Area (sqft) Area (m 2 ) (gpd) (m 3 /d) Rej (%) Year SWC SWC SWC SWC In recent years, there has been a focused effort by membrane manufacturers to increase membrane permeability and in turn, lower energy requirements. But higher permeability is not the only parameter by which to guage the performance of an RO membrane. Along with higher permeability is the need to maintain high rejection. The most successful membrane R&D programs have sought to increase permeability without sacrificing rejection. The

5 fruition of such efforts is a state-of-the art seawater element with a high flow of 9000 gpd and an average sodium chloride rejection of 99.8%. To better illustrate the combined high flow and high rejection of the latest seawater membrane, Figure 4 plots the water transport coefficient and salt transport coefficient of the SWC5 along with other seawater membranes. These coefficients proved a more accurate view of the membrane performance by decoupling the flow and salt passage and allowing them to be considered independently of one another. The current market offers several energy saving membranes as illustrated by their high water transport. SWC5, Mem A3, and Mem B3 all have similar high permeability. But a notable difference arises when focusing on the salt transport of each energy saving membrane. An increase in salt transport accompanies the increase in permeability for Mem A3 and B3 leading to a coefficient that is 50% to 66% higher than the SWC5. The SWC5 achieves high permeability without sacrificing permeate quality. Membrane Permeability (ml/(cm2 * sec * atm)) 4.5E E E E E E E E E E+00 Comparison of Seawater Membrane Permeability and Salt Transport SWC3 SWC3+ SWC4+ SWC5 Energy Saving Membranes Product Mem A1 Mem A2 Mem A3 Water Transport Salt Transport Mem B1 MemB2 MemB3 5.0E E E E E E E E E-06 Figure 4. Comparison of Energy Saving Membranes water transport and salt transport with other membranes produced by the various manufactures. SWC5 achieves a high water transport while maintaining low salt transport. Cost savings is the clear implication of high flow and high rejection performance for seawater desalination. Up to 80% of a desalination plant s power usage comes from the high pressure feed pumps of the first pass (Wilf, 2000). Compared to seawater membrane technology from five years ago, the use of the latest high flow membranes could reduce the first pass pumping power by 0.21 kwh/m3 and reduce the overall water cost by more than 35%. With the introduction of these new, high flow seawater membranes comes a challenge similar to those faced when the high flow brackish membranes were introduced in the 1990 s (Wilf, 1997). The high flow seawater membrane leads to a significant flux imbalance through the pressure vessel with the lead element seeing a much higher flux than the tail element. To more evenly redistribute this flux, lower flow SWC4+ membranes can be used in lead positions of the vessel with the high flow SWC5 loaded in the tail positions. Membrane Salt Transport (cm / sec)

6 Not to be overlooked is the operational and capital cost savings derived from a high rejection membrane producing a low salinity permeate. To meet final product water requirements, seawater desalination plants typically require a second pass brackish RO system to treat a portion of the permeate coming from the first pass. A full second pass can add as much as 20% to the overall water cost of the plant. Relative to a 99.7% rejection seawater membrane, the low salinity permeate coming from a 99.8% rejecting first pass would require less processing and therefore a smaller second pass to deliver the same capacity with the same quality. As an example, consider a 47,000 m3./d plant operating on a 40,000 mg/l seawater at 50% recovery in the first pass and 90% recovery in the second pass. If the overall product water quality requirement is less than 300 mg/l, only 5% of the SWC5 product would require processing by a second pass. However, if the SWC5s were replaced with 99.7% rejection membranes, the amount of permeate to be processed would increase to 35%. The first pass capacity would increase by 3.6% and the second pass capacity would increase nine times. Targeting Specific Ions As the limitation of membrane permeability and element design improvements are approached, membrane manufactures turn to improvements in membrane rejection and membrane modifications which target the rejection of specific ions. The NANO-SS, a new high permeability nanofiltration membrane, selectively rejects divalent salts while readily passing monovalent salts. When treating a typical seawater feed, for example, the NANO-SS rejects 99.7% of the sulfate ions while rejecting only 18% of the chloride ions. The membrane also has a molecular weight cut-off of about 160 Daltons, making it useful for the recovery of certain organic components from salt solutions. A membrane developed to specifically reject the boron ion is the ESPA B. Boron is a limiting ion in seawater desalination and in the semiconductor industry because it readily passes through an RO membrane. Based on proven polyamide technology, the ESPA B operates at lower pressures like other low energy brackish elements, but with the highest boron rejection of comparable membranes. At an elevated ph of 10, the ESPA B can achieve a boron rejection of 96%. With a higher boron rejection than conventional brackish elements, the use of ESPA B elements results in fewer elements or less chemical addition in the second pass of seawater desalination system. The use of ESPA B will also reduce the load on boron selective ion exchange downstream of an RO. Another advancement in selective rejection membrane technology is the low-fouling ESNA1- LF nanofiltration softening membrane. The ESNA-LF membrane is currently used by several municipalities in Florida, including Deerfield Beach, which draw a portion of their municipal supply from the Biscayne Aquifer. This shallow, highly rechargeable aquifer covers 4,000 square miles, ranges in depth from 0 to 240 feet and contains a high content of natural organic matter in the range of 8 to 20 mg/l total organic carbon (TOC). Treatment of the aquifer is required to provide drinking water of suitable quality. The average hardness is 235 mg/l as CaCO3. A level of less than 90 mg/l as CaCO3 is desirable in most softened waters. The high organic level in the aquifer imparts color in the range of 30 to 70 color units. A level of less than 15 color units is desirable in treated water. At Deerfield Beach an existing lime softening process has been used to meet these finished drinking water goals. However, the high concentration of chlorine required to lower the color to acceptable levels also converts a portion of the natural organic material to unwanted by-products, known as trihalomethanes and haloacetic acids. The US EPA has set legally enforceable standards for these two carcinogens. The limits are mg/l and mg/l for trihalomethanes and haloacetic

7 acids, respectively. The feedwater has an average trihalomethane formation potential of 400 mg/l and an average haloacetic acid formation potential of 300 mg/l. To meet their growing need for drinking water and to meet current and soon to be implemented regulated limits, the city of Deerfield Beach opted for a nanofiltration membrane process, also know as membrane softening, over expansion of the existing lime softening process. The permeate from the nanofiltration plant is blended with the lime softened effluent therefore the nanofiltration permeate limits were set lower than those listed above. The separation objectives of the nanofiltration process are to reduce the feed hardness from 235 mg/l CaCO3 to a range of 20 to 33 mg/l CaCO3, to reduce natural organic matter so that the permeate contains less than 2 color units, and to achieve a total trihalomethane formation potential less than mg/l and a total haloacetic acid formation potential less than mg/l. The section of the Biscayne Aquifer underlying Deerfield Beach also contains an average of 1.5 mg/l of iron. An additional objective is to reduce the iron to less than 0.20 mg/l. Un-aerated, ferrous iron is rejected at the same rate as the hardness ions of calcium and magnesium. (Figure 1) To be economically viable the nanofiltration units must operate at a transmembrane pressure of less than 90 psi and at a recovery of 85%. Transmembrane pressure is defined as the feed pressure minus the final permeate pressure and the recovery is calculated as the percentage of the permeate flow to the feed flow. Also the membrane must possess a resistance to fouling due to adsorption of natural organic material on to the membrane surface. At 85% recovery, the concentration of organics in the nanofiltration concentrate is approximately 6.5 times the concentration in the feed. To meet the demands of this separation objective, Hydranautics developed a new class of fouling resistant, selective rejection nanofiltration membranes. These new membranes, labeled the ESNA1-LF series, can be adjusted in the manufacturing process to meet the specific hardness passage requirements of individual water utilities. They all have a high rejection rate of natural organic material and exhibit a resistance to organic adsorption fouling. The 10.5 million gallon per day Deerfield Beach nanofiltration system consists of five units, each with a capacity of million gallons per day (10000 m 3 /day). Only four units are required to meet the total demand with the fifth unit in a standby mode. Each unit operates at 85% recovery and is a two-stage array of pressure vessels. Each pressure vessel contains seven spiral wound nanofiltration elements for a total of 504 each ESNA1-LF1 nanofiltration elements per unit. A total of 2,520 elements were supplied for the project. The units operate at an average flux of 13 gallons per square foot of membrane per day (gfd). The pretreatment to the nanofiltration system consists of sulfuric acid addition, scale inhibitor injection, followed by 5 micron cartridge filtration. This is the typical pretreatment to almost all of the reverse osmosis and nanofiltration systems in Florida that operate on un-aerated well water supplies. The permeate is sent to a degassifier for carbon dioxide reduction and later blended with lime softened effluent. The nanofiltration concentrate is directed to a pressurized sanitary main. All five units were started in November and December of All five units were within the specified ranges at startup and continue to meet specifications 2 years after startup. The system hardness rejection averages 95% and the total trihalomethane and haloacetic formation potentials of the permeate are well below limits. The average flux decline of all 5 units is approximately 10% after 2 years of service. To date, no unit has required a cleaning based on flux decline even though the feed TOC levels have been measured as high as 36 mg/l. In addition to the successful startup of this plant, the 40 million gallon per day Boca Raton nanofiltration plant, the largest in the world and located just 4 miles north of Deerfield Beach,

8 was brought on line during the period of August 2004 through April This plant utilizes ESNA1-LF2 membrane, a looser nanofiltration membrane tailored for higher hardness passage than the ESNA1-LF. This was possible due to the absence of iron at the Boca Raton location. In conclusion, the new ESNA1-LF has proven to be a low fouling nanofiltration with stable operating performance and hardness rejection that can be tailored as needed. Conclusion In summary, it can be seen that a broad variety of high performance products are available. Many of these products come from incremental advances in existing membrane and element technology. When combined, these advances can lead to significant capital and operational savings. RO system designers can take advantage of these advancements to optimize the system and reduce the cost of water treatment. Cooperative participation with the membrane supplier is often helpful to ensure that the products are designed and used in the most effective manner. References Bureau of Reclamation, Industry Consortium Analysis on Large RO and NF Element Diameters, Desalination and Water Purification Research and Development Report No. 114, Wilf, M, and Bartels, C., Optimization of Seawater RO Systems Design, Desalination 173, (2005) Wilf, M. Effect of New Generation of Low Pressure, High Salt Rejection Membranes on Power Consumption of RO Systems, AWWA Membrane Technology Conference, New Orleans, LA, 1997.