Reverse osmosis (RO) plays a vital role in most powerplant water treatment systems today, particularly those serving gas-turbine-based generating facilities (Fig 1). It is used in the production of water for power enhancement (evaporative coolers, fogging) and emissions reduction, makeup for heat-recovery steam generators, treatment of plant wastewater, etc.
Proper membrane care is critical to producing the quantity of permeate required at the quality necessary over the expected lifetime. The editors approached Rob Goodlett of Avista Technologies Inc, San Marcos, Calif, shortly after his presentation at the 2009 Southwest Chemistry Workshop to learn what RO-related tasks GT owner/operators should be factoring into their outage plans.
But before getting into details, short backgrounders on both organizations are useful:
- Avista Technologies’ core business is the supply of specialty chemi¬cals and technical support services for membrane-based filtration and RO systems.
- Southwest Chemistry Workshop is an annual meeting serving the generation sector of the electric-power industry. It focuses primarily on water treatment issues challenging owner/operators of plants powered by fossil and nuclear fuels. The 2010 conference, hosted by Capital Power USA, will be held June 8-10 in San Diego. Your contact is Frank Spencer (fspencer@capitalpowerusa).
Goodlett began by saying that the only things any owner/operator should have to do for the RO system during an outage are membrane replacement or offsite cleaning; everything else should have been done beforehand.
It’s important to monitor RO system performance daily, he continued, and to clean membranes and conduct other maintenance as operating data indicate the need. Failure to address issues in timely fashion can have a permanent adverse effect on system performance and/or reduce membrane life.
All pressure-driven membrane separation systems foul over time (Fig 2). The symptoms of fouling include a decrease in normalized permeate flow, an increase in salt passage, and an increase in differential pressure across the elements. RO element manufacturers recommend cleaning when permeate flow drops by 10% to 15% or when there is a 10% to 15% increase in differential pressure across the elements. Performance monitoring lets operators know when these limits have been reached.
Service companies like Avista help plant O&M personnel measure performance with proprietary normalization software. For example, Avista’s normRO, provided free to customers, considers changes in water temperature, pressure, and conductivity to differentiate between membrane fouling and changes in feedwater conditions. Data, recorded by operators on daily rounds, are entered into the program, thereby enabling performance trending so maintenance can be scheduled in advance.
Outage preparations should include an analytical review of current operating conditions. Goodlett also suggested an autopsy or cleaning study to determine membrane health and the degree of fouling, and to select site-specific membrane cleaners (Fig 3).
Success in the lab equates to success in the field, he added. Confidence in the cleaning formulation is particularly important because of the limited time available during an outage to restore membranes to as close to original condition as is possible. The cleaning process must meet expectations on first use.
Fig 4 illustrates the importance of effective cleaning. The red line shows that proper cleaning of a healthy membrane effectively restores production to the original baseline value each time. By contrast, ineffective cleanings, represented by the green line, restore less and less of the original flow each time. Note the similarity between these curves and ones you’ve seen previously to illustrate the benefits of compressor cleaning.
Additionally, when cleanings are not effective, you must clean more frequently. This increases downtime and operating costs, and shortens membrane life. Note that ineffective cleaning can be caused by poor procedures, the wrong formulation, and cleaning intervals that are too long.
Ensuring that your RO system is at full capacity and running well following an outage is critical. The last thing you want to deal with when bringing units back into service is interruptions in process water supply. If you don’t have personnel with sufficient RO experience onsite, consider hiring a qualified service organization to review daily processes and data collected, and to recommend a suitable maintenance program.
Goodlett told the editors there were five types of foulants to address when preparing a cleaning formulation: metals, scale, silt, organics, and chemical (Fig 5). He provided the following descriptive material:
Metal foulants most frequently include iron, manganese, and aluminum; sometimes zinc, copper, and nickel are found. Iron and manganese commonly occur in groundwaters as soluble divalent ions (Fe++, Mn++). Should air or chlorine be introduced into these waters, the iron and manganese may be oxidized and precipitate on membrane surfaces as hydrous oxides. These compounds include oxides (such as MnO) and hydroxides (such as Mn(OH)2).
Other sources of iron fouling: (1) Corrosion processes that occur within well headers or feed piping and (2) carryover from flocculator/clarifiers or multimedia filters when iron salts are used as coagulants. Fouling by aluminum oxide is common. The usual source is carryover of flocculated solids from filtration processes using aluminum salts as coagulants.
Scale. Supersaturation of sparingly soluble salts (those characterized by low solubilities) can cause scaling of membrane surfaces and feed spacers. The most common scale-formers are calcium carbonate and sulfates of calcium, barium, and strontium (Fig 6); less common scales are silica and calcium fluoride. Silica scale, when it does occur, is very troublesome because of the difficulty in removing it from membrane elements without damaging them. Sidebar 1 offers more detail.
Goodlett pointed out that scale often is found on the downstream end of the last membrane in the system, or on the inside of the last pressure vessel. Higher product-water conductivity may be your first indicator of scaling. He stressed that scale is important to eliminate as quickly as possible because it is abrasive and may harm the membrane surface.
Colloids and silt are the most common foulants. Examples include clays, colloidal silica, rust particles, and bacteria, which are found in virtually all surface water supplies (Fig 7). Colloids are particles less than about 1 micron in diameter and do not settle from a standing solution. Silt particles have larger diameters and will settle from solution. Indicators of colloidal fouling, Goodlett said, include high delta P across the first array and reduced permeate production.
Organics. Humic and fulvic acids frequently are found in surface waters, formed during the decay of leaves and other vegetation. The degree of fouling these acids cause depends on their exact nature and feedwater ionic composition. Today’s polyamide membranes are more susceptible to severe fouling than older cellulose acetate membranes.
Two more facts to remember: (1) Acids with high molecular weights are more serious foulants than those of lower weight. (2) Calcium and magnesium ions may contribute to humic and fulvic acid fouling by binding together anionic groups of membranes and acids.
Biologically derived organic foulants include the slime exuded from bacteria and filamentous fungi, as well as the microorganisms themselves. Slime may originate in the feedwater and be carried downstream to your membrane separation system. But more frequently it forms insitu through the growth of microorganisms on membrane surfaces (Fig 8).
Goodlett said biological slime is easy to identify in RO vessels and inlet vessels by sight and smell. Another telltale clue is high differential pressure across the first array. Biocides for mitigating slime formation are covered in Sidebar 2.
Chemical fouling generally is caused by feeding two or more incompatible chemicals ahead of the RO system. The most common example is the precipitation of polymeric antiscalants by organic coagulants which typically plug the front ends of first-array membranes. High differential pressure across the primary array is an indicator.
Plant owner/operators can clean their membrane systems with generic chemicals or special commercial formulations. The former, which include citric acid and trisodium phosphate, have a cost advantage. However, Goodlett points out, a selection based solely on cost may actually turn out more expensive for one or more of these reasons:
1. Options for controlling scale
Without some means of scale inhibition, sparingly soluble salts (slightly soluble in layman’s terms) will precipitate and solidify on RO membranes and in the flow passages of individual elements. Common scales include calcium carbonate and sulfates of calcium, barium, and strontium; less common are calcium fluoride and silica.
The effect of scale on the permeation rate of RO systems (Fig A) is characterized by an induction period when flow decreases rapidly. Length of the induction period varies with the type of scale and the degree of supersaturation of the sparingly soluble salt. Note that the induction period for calcium carbonate is much shorter than that for sulfate scales.
While there are effective cleaners for scale, economics strongly favor preventing its formation. One reason: Scale often plugs RO element feed passages, making cleaning both difficult and time consuming. There also is the risk that scaling will damage membrane surfaces.
Methods of control
Scaling typically is controlled by one of these three methods: acidification, zeolite softening, or antiscalant addition. Here’s what you should know about each:
Addition of acid destroys carbonate ions, removing one of the reactants necessary for calcium carbonate precipitation. But it is ineffective for preventing other types of scale, such as silica and some sulfates. Keep in mind that use of sulfuric acid to combat calcium carbonate precipitation contributes sulfate ions and increases the likelihood of sulfate scale. Hydrochloric acid avoids this problem, but it is more expensive.
Other disadvantages: (1) The potential for acid corrosion, (2) cost of acid tanks and metering/monitoring equipment, and (3) decrease in pH of RO permeate and increase in permeate conductivity. Regarding the last point, excess carbon dioxide contained in the permeate of acid-fed systems will increase the cost of ion-exchange regeneration unless removed by a degasifier.
Zeolite softening uses the sodium form of strong-acid cation exchange resin. Sodium in the resin is exchanged for magnesium and calcium ions in the RO feedwater. Specifically, Ca++ (Mg++) + 2NaZ ? 2Na+ + CaZ2 (MgZ2). When all sodium ions have been replaced by calcium and magnesium, the resin must be regenerated with a brine (sodium chloride) solution.
A benefit of zeolite softening is that it eliminates the need for a continuous feed of acid or antiscalant—unless silica scale is a potential (see below). In addition, some experts say it helps remove trace organics from RO feed streams and makes colloidal contaminants less likely to foul membrane surfaces. Though others agree that while resin adsorbs many different organic molecules and softening increases the negative electrostatic charge of colloidal solids, there is no persuasive evidence that either of these “benefits” significantly improves RO system performance.
The main disadvantage of softening, compared to acid or antiscalant addition, is cost. Avista Technologies’ Rob Goodlett said an economic analysis by his company for an RO system designed to produce 75 gpm of permeate at 75% recovery from feed waters with a wide range hardness values showed antiscalant significantly less costly than softening across the board.
One reason is that the cost of equipment for softening ranges from 10 to 20 times that for antiscalant addition. Also, while the cost of antiscalant remains relatively flat across the hardness range, the cost of salt increases dramatically. A present-worth comparison of the two systems showed antiscalant at x and softening at 4x for 10-ppm hardness. At 250 ppm, the numbers were 1.15x and 28x. Note that the foregoing comparison does not include disposal cost of spent softener regenerant, which can be significant in some areas.
Antiscalants are surface-active materials that interfere with precipitation reactions in these three ways: threshold inhibition, crystal modification, and dispersion. The most capable suppliers combine different antiscalants to accentuate one or more of these mechanisms to meet specific user needs.
Threshold inhibition is the ability of an antiscalant to keep supersaturated solutions of sparingly soluble salts in solution. As crystals begin to form at the submicroscopic level, negative groups located on the antiscalant molecule attack the positive charges on scale nuclei, interrupting the charge balance necessary to propagate crystal growth (Fig B).
Crystal modification is the property of an antiscalant to distort crystal shapes, resulting in soft, non-adherent scale. Scales treated with crystal modifiers appear distorted, generally oval in shape, and less compact.
Dispersancy is the ability of some antiscalants to adsorb on crystals or colloidal particles and impart a high anionic charge, which tends to keep the crystals separated (Fig C). The high anionic charge also separates particles from fixed anionic charges present on the membrane surface.
- Commercial cleaning formulations typically are more effective than generic chemicals, so you need less of them.
- It takes time to mix generic cleaners.
- Generics can damage membrane element components if a high or low pH excursion occurs (Fig 9).
Good commercial formulations are more effective than generic chemicals, he explained, because they contain a “team” of ingredients that work together to remove foulants. The general categories of ingredients are bulleted below to help you understand what functions they perform.
- Buffers maintain the cleaning solution’s pH in the range necessary for the formulation’s constituents to work at maximum efficiency, as well as to protect membranes and other RO system components.
- Surfactants “wet” surfaces and enable the cleaning formulation’s other components to penetrate layers of foulant.
- Builders are complex divalent ions—such as calcium and magnesium—that may render surfactants inactive.
- Chelants sequester heavy metals—such as iron and manganese—as well as the constituents of many scales.
- Dispersants disperse colloids and silt components that cannot be dissolved by the cleaning formulation. They work in concert with surfactants.
- Hydrotopes are organic substances added to some formulations to dissolve targeted organics—such as bacterial slime.
- Redox controllers raise or lower the oxidization potential of the cleaning solution. To illustrate:
2. Deal quickly, effectively with biofouling
Polyamide membranes have several advantages over the older cellulose acetate membranes for RO systems. They (1) improve salt rejection, (2) resist hydrolysis, (3) resist bacterial degradation, and (4) tolerate cleaning formulations. The main disadvantage of polyamide membranes is their intolerance to chlorine and other oxidants.
Keep in mind that without some form of disinfection, microorganisms may quickly colonize and foul membrane surfaces and plug element feed passages. Left unchecked, the effects of biological fouling become increasing irreversible, and the increase in element differential pressure may result in secondary mechanical deformation of individual elements.
The cumulative effects of biofouling include higher cleaning and maintenance costs, lower-quality product water, and significantly reduced element life. Figure illustrates the effect of severe biofouling on RO system flow: Following a period of decline, flow stabilizes at a reduced equilibrium value-a pattern similar to that seen with colloidal fouling.
Many bacteria that contact membrane surfaces adhere irreversibly to them, multiply, and exude foulants. Researchers have shown that the adhesion process is very rapid. Also, RO membrane surfaces have a finite number of bacterial attachment sites, which when completely filled, become unavailable for further adhesion.
Many species of bacterial, filamentous fungi, and yeast have been associated with biofouling, which often is accompanied by mineral deposition-including scale. Inorganic ions identified in biofilms generally are insoluble-such as calcium, aluminum, iron, sulfur, phosphorus, and silicon.
Biofilms consist of a complex mixture of microorganisms, the byproducts of microbial metabolism, and inorganic salts. Experience indicates that second-stage elements generally contain larger quantities of biofoulants than first-stage membranes.
Control of biofouling is by injection of non-oxidizing biocides into RO feed streams, either continually or intermittently. Some of these chemicals are poorly rejected by RO membranes. If they cannot be tolerated in the product, an alternative solution is required. Options include biocides that are completely rejected by the membrane, or those that kill microorganisms quickly enough that they can be injected while the system is offline.
Another alternative: Ultraviolet destruct systems downstream of the RO modules. They are used in the semiconductor industry to convert organic compounds-including biocides-into simple charged particles, which can be removed by ion exchange. In such systems, passage of biocide into the product is less of an issue than it may be in powerplants.
The choice of which biocide to use depends on user requirements and the type and number of bacteria present. Service providers, such as Avista, offer laboratory screening tests of their biocides. All the user must do is collect a system feedwater and concentrate sample in a sterile, insulated container and send it in for analysis.
Begin treatment immediately after identifying the most effective biocide. Initial dose typically is high, to “shock” the system. The dose rate is reduced as biofouling control is achieved. Control is indicated by a marked decrease in the rate of decline in normalized flow and differential pressure. Bacterial counts are another indicator. When the numbers of bacteria present in the concentrate stream equal those of the feed-taking the concentration factor into account-control is achieved.
For faster results, consider thoroughly cleaning your RO system before beginning biocide treatment-this to remove accumulated biofilm deposits on membrane surfaces and within element flow passages.
Finally, when feasible, clean and sanitize feed piping. If this is not done, large quantities of accumulated biofilm may slough off piping walls and plug cartridge filters and feed-end RO elements when biocide treatment is initiated.
Reducing agents help dissolve iron and manganese; mild oxidizing agents help remove organic foulants.
Goodlett told the editors that when users find it difficult, or virtually impossible, to get their membranes clean with a given formulation—possibly one that has been effective in the past—a cleaner evaluation study should be conducted. Foulants tend to occur in combinations based upon the feed source and the physical and chemical nature of the water supply often changes over time. It’s not unusual to find gremlins in your makeup water.
Typically, one membrane element from the affected system is removed and samples with representative fouling are removed. They are wet-tested and cleaned with several candidate formulations. At Avista, this is done in the company’s RO Vision™ test cell apparatus.
Refer back to Fig 3 for pictures of test cells, which have transparent tops that allow technicians to observe the cleaning and collect membrane performance data. A foulant analysis also may be conducted as part of the study. Knowing the chemical nature of the foulants helps chemists optimize the pretreatment process.
For example, when fouling is heavy and/or tenacious, it sometimes is necessary to soak the system overnight in the appropriate cleaners. During the soaking period it often is helpful to occasionally “bump” the cleaning pump to agitate the solution within the membrane elements.
A suitable cleaning system for membrane separation systems is relatively simple and one that most plants might build in-house (Fig 10). Size the cleaning tank, which should be made of reinforced plastic or other non-corrosive material, to hold a volume equal to that of the water contained in the first-stage pressure vessels, cleaning-system hoses, and cartridge filter holder.
Some help here for your volumetric estimate: Pressure vessels for 4-in.-diam × 40-in.-long membrane elements hold 2 gallons per element; 6-in. elements, 3.5 gal; 8-in. elements, 6 gal. The volume of liquid retained per foot of 2-in.-diam hose is 0.16 gal; 3 in., 0.37 gal; 4 in., 0.65 gal.
Equip the tank with a cover, heater, and temperature and level controls. Include an exhaust fan in your design if the tank will not be located in a well-ventilated area. Specify a centrifugal cleaning-solution pump made of Type-316 stainless steel to deliver 10 gpm to pressure vessels retaining 4-in. membrane elements (23 gpm for 6-in. elements, 40 for 8 in.).
Be sure the hoses that connect the cleaning skid to the RO system have an adequate pressure rating for the task, Goodlett cautioned, and that the hose is not kinked before operating. Also, locate the return line below the liquid line in the tank to reduce splashing and foaming.
Other best practices he recommended implementing:
- Install a 5-micron cartridge filter after the cleaning pump to remove undissolved powder cleaners.
- Locate a safety strainer upstream of the cleaning pump.
- Provide appropriate valves to control system pressure and permit drainage and flushing of tank and lines.
- Install a flowmeter to establish the proper rates through the pressure vessels.
- Provide a bypass line around the cleaning loop and back to the tank to help control flow to the pressure vessels and to mix or dissolve cleaning formulations.
Good procedures are essential for effective cleaning. Here are key points to remember:
Flow and pressure. Clean one stage at a time, circulating the cleaning solution through each stage for at least one hour. Flow should be at the maximum rate recommended by the membrane element manufacturer. If unknown, use the flow rates suggested for system design presented in the preceding section.
Clean at the minimum pressure needed to achieve the desired flow. Low pressure minimizes permeation and reduces the fluid force that holds foulants to the membrane surface. In general, do not exceed 60 psig.
Dilution water used in the preparation of cleaning solutions should be RO permeate or demineralized water.
Flush the system with RO permeate or demineralized water prior to and after cleaning. If two formulations are used sequentially, flush the system between cleanings.
Temperature of the cleaning solution should be at the maximum permitted by the element manufacturer. For thin-film polyamide elements, Goodlett considered 120F both safe and effective for solutions having a pH value of less than 10.5; for cellulose acetate he said not to exceed 95F.
Cleaner quantity. Be sure to calculate the total volume of water in the cleaning loop to determine how much of the cleaning formulation is required, he continued. This includes the amount of water in the cleaning tank, pressure vessels, hoses, and cartridge-filter housing. The numbers presented in the previous section will help here.
To give you a feel for the amount of water you’re dealing with, Goodlett ran through a quick calculation for a system with a 4:2 array having six 8-in.-diam × 40-in.-long elements per vessel. His sample system had 200 ft of 3-in. hose to connect the cleaning skid with the RO system, the cartridge filter housing held 15 gal, and the cleaning tank volume equaled the volume of the hoses, pressure vessels, and filter housing.
Total volume required to clean the first stage: 466 gal. The second stage needed a minimum volume of 322 gal. With this information, together with the cleaner concentration recommended by laboratory tests, you can determine the quantity of the cleaning formulation required.
Goodlett noted that you can repurpose the cleaning solution used in the first stage for the second stage provided the pH is still in the recommended range and the solution is not turbid. ccj