B. Buecker, Buecker & Associates, Lawrence, Kansas (U.S.)
The refinery, petrochemical, chemical and other manufacturing industries have seen and continue to see many retirements of senior technical employees, some of whom have been with their plants since startup. New personnel are often confronted with challenges that require knowledgeable and quick action, which they may have difficulty providing without thorough training. A very important—and sometimes overlooked—unit operation within these facilities is makeup water treatment for steam generators and other processes. Many cases have been documented in which makeup system upsets have led to boiler tube failures that partially or completely shut down unit processes. Note: Cooling water systems are also often neglected to the detriment of plant operations. Parts 1 and 2 (September 2024) of this article serve as a guide for those employees who may be on a steep learning curve when it comes to makeup system understanding. Part 3 (October 2024) will conclude this series.
Membrane technologies—Serving both pretreatment and high-purity applications. Water purification by membrane filtration has expanded dramatically in the last four decades. Well-known, of course, is reverse osmosis (RO), but other membrane technologies have emerged for pretreatment. The most prominent are micro- and ultrafiltration (MF and UF, respectively). The filtering capacities of the various technologies are shown in FIG. 11.
For fresh water sources that do not encounter extreme fluctuations in suspended solids, MF and UF may be an excellent alternative to a clarifier. FIG. 12 illustrates a 300-gpm MF unit that replaced an aging clarifier.
This system produced, and still produces, effluent with a turbidity of < 0.05 nephelometric turbidity units (NTUs). The MF feeds a downstream RO unit with ion exchange effluent polisher that provides makeup for a supercritical power boiler. A cutaway view of the spaghetti-like hollow fiber membranes is shown in FIG. 13.
The process can be thought of as a combination of crossflow and dead-end filtration. Water flows along the length of and percolates through the membranes, while the suspended solids collect on the outer surfaces. Note: Some designs have an inside-out membrane flow path.
The pressure differential across the membranes is known as trans membrane pressure (TMP). Modern MF and UF units are fully automated, and when the TMP increases to a certain set point, an automatic backwash begins (FIG. 14). Backwashes are usually accompanied by air scouring.
Standard for modern units is a feature known as chemically-enhanced backwash (CEB). At a regular frequency—for example, every eight backwash cycles—cleaning chemicals are injected into the backwash water. Bleach/caustic will attack microbiological deposits, while citric acid loosens iron oxide particulates.
Even regular backwashes do not remove all particulates, and over time, the TMP will increase.
A periodic manual cleaning (e.g., every 2 mos–3 mos) is required to prevent irreversible fouling. Bleach and citric acid are common choices, as MF and UF fibers are durable and can withstand strong solutions. An important point to note is that membrane durability allows for continuous oxidizing biocide feed, if desired. This arrangement provides constant control of micro-organisms.
RO. As power and industrial plants increased in size and complexity in the last century, the need for high-purity makeup water became acute. A dramatic solution to these needs was the development of synthetic ion exchange (IX) resins that, when configured correctly, can reduce makeup dissolved solids concentrations to low part-per-billion levels. For low-pressure steam generators, basic sodium softening often provides the needed makeup purity, which is addressed shortly. However, a key point that has influenced makeup water treatment system evolution is that standalone IX units have limited capacity and require regular resin regenerations. The regeneration process consumes chemicals and manpower. Accordingly, RO has emerged as the core dissolved ion removal process in many high-purity makeup systems. FIG. 15 illustrates the principles of osmosis and RO, respectively.
Osmosis is a natural process. When two solutions of different ionic concentration are separated by a semi-permeable membrane that inhibits ion passage, water will flow from the dilute solution to the concentrated solution to establish equilibrium. The water movement exerts pressure on the membrane. In RO, external pressure is applied to the concentrated solution to generate purified water.
Modern RO systems are of the spiral-wound design in which the membrane along with spacer sheets are wrapped around a perforated core. Each assembly is known as an “element,” as shown in FIG. 16.
For most RO systems, the element dimensions are 8 in. (diameter) by 40 in. (length). The surface area of a single membrane has increased from an original standard of 365 ft2 to 400 ft2 or even greater. Multiple elements (typically five or six per pressure vessel) are assembled in series (FIG. 17), with multiple pressure vessels needed for normal industrial applications.
Per FIG. 16, feed enters the front end of each element and passes along the feedwater carrier, where the feed pressure forces water through the membrane (FIG. 18). The purified water (permeate) flows to the central core, while the increasingly concentrated feedwater (concentrate or reject) exits the element. Brine seals prevent feedwater from short-circuiting the elements. Anti-telescoping devices prevent the water pressure from pushing the membrane and spacer sheets out of the element. A typical RO pressure vessel will have five or six elements.
This configuration purifies water via the mechanism known as crossflow filtration.
A key aspect of crossflow filtration is that the reject impurity concentration continually increases as the feed passes from inlet to outlet. This concentration increase can have a large impact on membrane performance and selection of scale control chemistry, as will be discussed. A basic RO configuration is shown in FIG. 19.
For normal surface and ground waters, each stage will produce approximately 50% purified water (permeate) and 50% reject. Thus, the overall efficiency of a standard two-stage RO is 75%. Modern RO membranes can remove > 99% of dissolved ions.
A very important concept in RO design is the flux rate. Each element can pass a certain amount of water, and this volume is usually measured in gallons per day (gpd). Common values for 8-in. x 40-in. elements range from 4,000 to > 13,000 gpd. The rate at which water passes through the membrane is known as the flux and is measured in gallons per square foot per day (gal/ft2/d). The general purity of the water partially dictates the flux rate. General guidelines suggest:2
To illustrate the calculations that go into sizing an RO system, consider a two-stage unit designed to produce 300 gpm of purified water with 75% recovery of the feed. Conditions are as follows:
Because 300 gpm of permeate is required, and the overall efficiency is 75%, the influent flowrate must be 400 gpm. This equates to 576,000 gpd. At a capacity of 10,000 gpd per element, 58 first-stage elements are required. Rounding off gives 12 pressure vessels with 5 elements apiece in the first stage. Because 50%, or 288,000 gpd, of the first stage is concentrate, the theoretical number of elements required in the second stage is 29. Rounding off gives 6 pressure vessels with 5 elements apiece.
If pristine water serves as the makeup, it is sometimes possible to treat the second-stage reject in a third stage to give an overall RO output of 87.5%, but this may require enhanced chemical treatment. In that regard, high-production RO systems have evolved that, with special chemistry treatment, can potentially operate at 90% recovery or greater. High recovery RO units are finding application in the industrial wastewater treatment industry, sometimes as concentrating mechanisms for zero liquid discharge (ZLD) systems. The author assisted on one such project at a combined-cycle power plant in the northwestern U.S.
For high-pressure steam generators and other high-purity water applications, the two-pass RO design shown in FIG. 20 is common.
In the two-pass configuration, the permeate from the first pass is further treated in the second pass to produce water of even higher purity. Note that the second-pass reject is of good quality and can be recycled to the RO inlet rather than discharged to waste.
RO feed and process treatment needs. RO membranes are subject to a variety of fouling/scaling mechanisms, which are influenced by the membrane location (first or second stage). First-stage membranes are most susceptible to particulate buildups and fouling by large organic molecules, microbes and possibly excess carryover of clarifier coagulants. Scale formation is the primary concern in second-stage membranes.
The importance of pretreatment to reduce suspended solids loading has already been noted, but further discussion is necessary. Particles that escape upstream treatment accumulate in the leading RO membranes. RO units typically are equipped with 10-, 5- or 1-micron cartridge filters as a last line of defense against particulate ingress. FIG. 21 illustrates heavy particulate loading on a set of filters.
Particulate accumulation increases the pressure differential across the filter assembly, which can, in turn, influence the flow and pressure to the RO feed pump.
An important measurement for determining particulate fouling potential is the silt density index (SDI). SDI is a spot test, where a flowing sample is passed through a 0.45-µm filter at 30 psig pressure. Measurements are made of the time required for 500 milliliters (ml) of water to pass through the filter at the beginning of the test (ti ) and again after 15 minutes (tf ).
The SDI is calculated using Eq. 10:
SDI15 = (1 – (ti / tf )) / T × 100 (10)
As an example, consider the following data collected from an operating RO unit:
ti = 34 sec
tf = 66 sec
T = 15 min
Per Eq. 10, the SDI15 = 3.2.
A general rule of thumb is that SDI should be < 5 and preferably < 3 to protect RO membranes. However, SDI is not necessarily the only criterion to determine potential particulate fouling. The type of water and/or the nature of contaminants are also important considerations. The author is aware of one case where the SDI of the RO feed always ranged between 1 and 3, yet the membranes fouled with very fine iron oxide particles.
Other agents that can cause fouling or membrane damage are shown in TABLE 3.
Simple logic suggests that oil and grease will rapidly foul membranes, thus the recommended low limit. Large organic molecules (including natural compounds such as tannins and lignin from vegetation) and synthetic organics (should they leak from a process stream into the makeup) can coat membranes. Dissolved metals can be problematic. Of particular note is aluminum, which can potentially appear as coagulant carryover from an upstream clarifier. Coagulants have a cationic charge and will bond irreversibly with RO membranes, which have a slight negative charge.
As the process water passes along the RO membranes and purified water is extracted, the concentration of dissolved minerals in the reject increases (FIG. 22). For a single-pass two-stage RO, the increase is four-fold at the final second-stage membranes. Carbonates, sulfates, silicates and hydroxide compounds constitute the most common deposits. Accurate water analyses and consultation with the RO vendor are important to select the proper scale inhibitor treatment program. Dosages may need periodic adjustment to follow changes in feed water chemistry.
At this point, a review of recommended RO instrumentation is warranted to identify other factors that influence RO performance, including those related to microbiological fouling and scale formation.
SDI has been outlined above, and the next section goes into more detail about microbiological fouling. The following list outlines in brief the other important monitoring items from FIG. 22, moving from inlet to outlet.
Microbiological control. Cartridge filters, and especially the narrow channels in an RO membrane element, serve as excellent locations for microbiological fouling (FIG. 23).
FIG. 22 shows two ports (RO feed and RO reject) to collect grab samples for microbiological analyses. Typical guidelines are [colony forming units (CFUs)]:
A balancing act is often required to protect membranes from microbiological fouling and biocide attack. Chlorine, typically injected as a bleach solution, has been the common choice for microbiological control in clarifiers and filters (and MF and UF systems) for many years. However, unreacted chlorine that continues along to the RO can severely damage membranes. Two methods are common to remove chlorine ahead of an RO: reducing agent feed (as shown in FIG. 22) or placing an activated carbon filter in the pretreatment system. However, some microorganisms go into hibernation when subjected to an oxidizing biocide, and then will re-emerge when the biocide is removed. A relatively convenient place to visually inspect for fouling is the cartridge filter housing (FIG. 24).
The question then arises of how to protect cartridge filters and RO membranes from organisms that come out of hibernation. Two methods are becoming increasingly popular:
Periodic CF visual inspection can also reveal other issues. For example, the presence of anthracite, sand or garnet in the filters suggests a broken lateral or nozzle in an upstream multi-media filter. Heavy iron oxide deposits on CF filters indicate upstream corrosion of carbon-steel piping or some similar source.
Following any cartridge filter inspection or filter change-out, it is important to thoroughly rinse the vessel to remove debris that might otherwise foul the filters or be washed to the RO membranes. Safety is important when inspecting CF vessels, especially if microbial colonies are present. Bacterial slime can harbor harmful organisms such as Legionella. Thoroughly scrubbing the vessel with a bleach solution will kill such organisms.
Normalization. RO normalization programs are always recommended to monitor performance. All reputable RO vendors have a normalization program, often in spreadsheet format. The program evaluates data from the instruments in FIG. 23 and provides an analysis of RO performance. Normalization is particularly important for identifying the need for membrane cleaning that otherwise might be masked by inlet water temperature changes. Even for RO units with excellent pretreatment systems and process chemistry programs, deposits will gradually accumulate on membranes. Without periodic deposit removal, irreversible fouling may occur (FIG. 25). Sometimes, RO units can operate for months without cleaning, but eventually will reach the rule-of-thumb 10% performance degradation that calls for cleaning.
When the normalization program indicates time for an RO cleaning, a multi-step process is employed:
It is important to individually clean each stage with a fresh solution.
Part 3 will appear in the October issue and will discuss ion exchange technology, including sodium softening, which is a common technique for industrial boiler makeup water production. HP
DISCLAIMER
The author has included much information in this article from consultation with Ed Sylvester, Director; Filtration, Ion Exchange & Membrane Technologies, ChemTreat Inc., Glen Allen, Virginia (U.S.), who focuses heavily on the refinery industry. However, the views represent those of the author. None of the programs or methods outlined in this series should be implemented without first contacting reputable manufacturers and consultants.
NOTES
LITERATURE CITED
Brad Buecker is President of Buecker & Associates LLC, and specializes in consulting and technical writing/marketing. Most recently, he served as a Senior Technical Publicist with ChemTreat Inc. Buecker has many years of experience in or supporting the power industry, much of it in steam generation chemistry, water treatment, air quality control, and results engineering positions with City Water, Light & Power (Springfield, Illinois, U.S.) and at the Kansas City Power & Light Company's (now Evergy) La Cygne, Kansas (U.S.) station. His work has also included 11 years with two engineering firms, Burns & McDonnell and Kiewit, and he spent two years as acting Water/wastewater Supervisor at a chemical plant. Buecker earned a BS degree in chemistry from Iowa State University with additional course work in fluid mechanics, energy and materials balances, and advanced inorganic chemistry. He has authored or co-authored > 250 articles for various technical trade magazines, and he has written three books on power plant chemistry and air pollution control. He is a member of the ACS, AIChE, AIST, ASME, AWT, CTI, the Electric Utility Chemistry Workshop planning committee, and is active with the International Water Conference and Power-Gen International.