New NERC standard aims to mitigate effects of GMDs

Special to CCJ ONsite by Thomas F Armistead, Consulting Editor

Here comes the sun, and your plant is in its cross-hairs. The solar energy that sustains life on this planet cycles through pulses of activ­ity roughly every 11 years. The puls­es are evident in a crescendo of sun­spots, flares, and other eruptions as the orb builds toward what astrono­mers call the solar maximum. Then the eruptions wane, returning in another dozen years or so.

When one of the eruptions is aimed directly at Earth, the results can be catastrophic. Loss of reactive power is the most likely outcome from a severe solar storm centered over North America, according to a report released in 2012 by the North American Electric Reliability Corp. Significant losses of reactive power could lead to voltage instability and, if not identified and managed appropriately, power-system voltage collapse could occur. In response to the findings, NERC is developing and gradually implementing new standards for the continent’s bulk-power system.

How bad could it be? On Mar 9, 1989, the Kitt Peak National Obser­vatory on the Tohono O’odham Reser­vation in Arizona reported a powerful flare on the sun. The next day, an explosion on the sun ejected a cloud of electrically charged particles 36 times as large as Earth directly toward our home planet. The cloud arrived at 2:44 a.m. (Eastern), March 13.

The wave of electrons and protons washed over and around Earth’s magnetic field, was channeled into the magnetic field lines, converged at the poles, and induced electric cur­rents at the higher latitudes. Where the ground was conductive, the geomagnetically induced cur­rents flowed harmlessly through the ground.

But the resistive igneous bedrock of the Canadian Shield forced the current to seek easier conductance. It entered the electric grid through ground wires and propa­gated throughout Hydro-Québec’s system. Within 92 seconds it brought down almost the entire grid, knock­ing 21,500 MW off line. The blackout lasted more than nine hours, affected 6 million people, and cost $2 bil­lion.

Quebec’s system was the most obvious victim of that event, but bulk-power systems throughout North America, including the US, experi­enced more than 200 related trans­former and relay problems. The most serious of those was the loss, from overheat­ing and permanent insulation dam­age, of a $12-million, 22-kV genera­tor step-up transformer at the Salem Nuclear Plant, Hancocks Bridge, NJ. Even with a spare transform­er available, the plant was offline for 40 days, entailing millions of dollars in lost power sales and replacement power purchases, in addition to the replacement cost of the transformer.

Playing defense. To forestall a repetition or worse outcome, in January 2015, NERC filed TPL-007-1, “Transmission System Planned Performance for Geomagnetic Disturbance Events.” The standard requires North American ISOs and utilities to perform state of-the-art vulnerability assessments of their systems and equipment for potential impacts from a severe, once-a-century benchmark geomagnetic disturbance event (GMD) and to mitigate identified impacts. Mitigation could include changes in system or equipment design or the installation of hardware to monitor or reduce the flow of geomagnetically induced currents (GIC).

TPL-007-1 was approved by the Federal Energy Regulatory Commission in September 2016.

The requirements are designed to be implemented over a five-year period, said Mark Olson, NERC senior engineer in reliability assessments. Entities began implementing the standard in 2017 and must take several steps leading to completion of the vulnerability assessments and mitigation plans by 2022. At FERC’s direction, NERC is developing a revision to be labeled TPL-007-2. It will enhance the benchmark GMD event used in the vulnerability assessments and is due for completion by May 2018.

Another NERC reliability standard, EOP-010-1, was approved by FERC in June 2014 and took effect in early 2015. The EOP-010-1 standard requires grid operators to have procedures that can be put in place to reduce impacts of severe GMD events.

Anticipation. The National Oceanic and Atmospheric Administration (NOAA), which operates the National Weather Service, has a Space Weather Prediction Center in Boulder, Colo, which monitors solar activity. And like the National Weather Service, which forecasts hurricanes by observing precursors such as tropical depressions in the Atlantic, the Space Weather Center tracks the development of sunspot groups to forecast GMDs, said Bill Murtagh, program coordinator for the center.

When large, complex sunspot groups emerge, “they may create eruptive activity called mass ejections that can have impacts here on Earth,” he said. “Isolated, complex sunspot clusters will produce coronal-mass ejections (CMEs) of significance that will create sizeable GMDs here on Earth.”

The current 11-year solar cycle began in December 2008 and reached solar maximum in April 2014, Murtagh said. “In the waning stages of the solar cycle, we get quite a few coronal-hole high-speed solar wind streams.” Coronal holes are large areas on the sun with lower magnetic fields allowing increased solar wind, he said.

“Solar-wind instruments will typically measure in backgrounds around 400 kilometers per second. One of the high-speed streams of solar wind associated with coronal holes will sweep past the Earth and buffet the Earth’s magnetic field for a couple of days, sometimes up to about 600 km/sec, and that’s a moderate-level storm.”

At this stage of the cycle, we should expect an increase in “moderate-level GMD, but we will continue to see isolated strong GMD associated with sunspot groups,” Murtagh said. “We typically see a couple of big clusters emerge over the course of these waning years at the three- to five- or six-year point.” A big sunspot group emerged in early September, he added.

The next solar maximum is expected in 2025, but Murtagh would not predict the intensity of the ramp-up to the maximum. “The 11-year cycle in itself is obviously quite predictable,” but the GMDs are not. He compares predicting GMDs with predicting hurricanes at the beginning of hurricane season. “Sometimes you get two or three hurricanes and other times you can get 20 hurricanes in a season. In the sunspot cycles, we’ve seen very big cycles, like Solar Cycle 19 back in the 1950s and early 1960s, and then we’ve seen this cycle, which is actually the smallest cycle since the first decade or two of the 1900s.”

Geomagnetically induced currents are the terrestrial events caused by GMDs. “When a coronal-mass ejection with its own magnetic field that gets shot out from the sun hits Earth, it hits Earth’s magnetic field, so we get two magnets coming together,” Murtagh said. “The interaction of the magnetic fields ends up inducing currents above us in the atmosphere, above the ionosphere, and those currents above manifest themselves on Earth in the form of geomagnetically induced currents.”

Earthly effects. Earth is affected only when a CME is directed at Earth, but there is no way, given the state of the science, to forecast when a CME directed at Earth will occur. Instead, Murtagh expresses the situation in probabilities. If he detects a very large sunspot cluster in the middle of the sun, “my probability of an eruption of a CME to occur is at 80%.” Probability is as much as he can do with the science he has. Being able to identify a pre-eruptive signature that would allow more precise forecasting is one of the industry’s “holy grails,” he said. “We’re not there yet.”

Like the National Weather Service, Murtagh produces routine daily forecasts. If sunspot activity indicates an 80% probability, NOAA would announce that level of probability, as a local weather station might announce a 50% chance of a thunderstorm. But when the threat reaches a certain threshold, he starts putting out alerts and warnings, just as a weather station might announce a tornado watch or tornado warning.

“We do it the same way. That product would go out as a forecast, saying, ‘80% chance of an eruption today.’ Then when the eruption occurs, and especially when we see a CME that’s Earth-directed, we issue a ‘watch product,’” which, like a weather watch, alerts recipients to the potential for an active threat.

NERC asked EPRI for assistance in responding to FERC’s request for validation of the GMD standard, said Rob Manning, EPRI’s VP of transmission and distribution infrastructure, and the research organization is working with the electric utility industry to evaluate the standard in further depth, doing calculations and validations and technical bases for the standard.

A geomagnetic disturbance created by a solar flare has only one characteristic: geomagnetically induced currents, primarily on the transmission system, Manning said. It manifests itself as heating in the transformer cores, depending on how high the current gets, because it saturates ac transformer cores. The primary danger from a solar-flare-type occurrence would be the loss of transformation in the switchyard. A secondary outcome would be harmonics in the plant. “If you’re protected against harmonics in general, you’re probably pretty well protected against GIC,” he said.

EPRI sponsors Sunburst, a system of sensors deployed around North America, which measures GICs on the grid and reports them to utilities. “We can tell you at any time what the actual GIC current flows are,” Manning said.  “With a solar flare, we generally believe we would have adequate time to isolate transformers that are potentially susceptible to damage.”

Unless you get a particularly catastrophic solar flare, it would be relatively slow heating, so you would have several minutes to several hours to take a transformer offline, which would then completely protect it from GIC.

The net result from that might be an unscheduled shutdown of the generating plant if there were a solar flare and the operators were able to confirm two things: one, that they are seeing GIC currents flow in the Sunburst system and two, the transformer is showing additional heating. If you see those two factors, that might be enough to shut down the plant to save it. You can reboot it back after the solar flare is past.”

Distributed energy resources, energy storage, and other latter-day evolutions in the grid have not materially modified the risks of exposure to GIC. “The primary susceptibility to GIC currents is very long lines, radial lines,” Manning said. A very long radial line can accumulate GIC, and it can only go to the end of that line, where it can damage equipment. Microgrids and distributed solar power are not likely to be susceptible to GIC.

At risk. Transformers are the generation infrastructure most at risk from a GMD, but other equipment also is exposed. “Potential effects include overheating of auxiliary transformers; improper operation of relays; heating of generator stators; and possible damage to shunt capacitors, static VAR (volt-ampere, reactive) compensators, and filters for high-voltage dc lines,” added an article in the spring 2011 issue of the EPRI Journal.

Powerplant operators have at least two main ways to protect their equip­ment from damage in a geomagnetic storm, said Buddy Dobbins, director of machinery breakdown in the Risk Engineering Dept of Zurich Services Corp, Schaumburg, Ill: They can take equipment offline when NOAA issues a warning of a CME or they can hard­en transformers against geomagnetically induced current.

Completion of Reliability Standard TPL-007-2 by NERC and approval of the standard by FERC by 2022 will provide transmission planners with new tools to help ensure the bulk-power system’s ability to withstand and mitigate the effects of potentially damaging solar activity.

Posted in CCUG | Comments Off on New NERC standard aims to mitigate effects of GMDs

Consider the impact of new operating regimes on your SCR

It’s easy to forget about the “big box” selective catalytic reduction (SCR) unit sandwiched in your HRSG modules, even though it stands between you and compliance with your air permit, and your ability to operate.

At the 2017 conference of the Combined Cycle Users Group (CCUG), held in Phoenix the last week of August, Andy Toback, Environex Inc, did his best to remind users that SCR process parameters have to be re-evaluated when gas turbines are upgraded, plant operating tempos change, duct-burner operation is more prevalent or variable, and/or the latest G and H technology machines are being deployed. Otherwise, you may be leaving money on the table or setting yourself up for unexpected costs and performance issues down the road.

Toback’s main message is that you need to adjust the original design expectations for the SCR based on real-world operating data, the key to optimizing the process for new conditions.

For an actual 7FB.01 to 7FB.04 upgrade, for example, the new combustors added 7 MW of output and lowered turbine NOx levels entering the SCR (Fig 1). Because the catalyst going forward typically will convert 9 ppm NOx levels from the turbine, compared to the original design of 25 ppm, the relative catalyst activity level, representing the expected end of life, can now be projected out beyond 10 years (Fig 2), compared to 5.5 years with the original combustors.

Ammonia consumption and ammonia slip at the stack also are reduced significantly because of the lower NOx conversion requirement for the SCR system. Due to the improved combustor dynamics, gas-turbine CO levels are not expected to change even though the NOx levels have decreased.

Toback’s second example is a 7FA.03 to 7FA.04 + DLN 2.6+ upgrade (Fig 3). The good news is that the megawatt output gain was higher than expected, another reason why actual measurements are important. However, peak-load fuel and exhaust flows increased accordingly, the turbine exit NOx levels did not change appreciably, but the SCR temperature increased by about 20 deg F after the upgrade.

The original Dot 03-machine catalyst design life projection was about 21 years. The upgrade design expectation reduced it to around 13 years. Based on the actual operating data, however, the catalyst life should be more like 18 years (Fig 4). Both an expected increase in engine NOx from 9 to 10.4 ppm and higher exhaust flows impair catalyst life, but because the actual operating data showed no change in gas-turbine NOx levels, the minimum catalyst-activity requirement did not increase as much as anticipated.

For advanced G and H machines, the news isn’t so good. Toback states, “We’re being asked to achieve the same 2 ppm NOx and ammonia-slip levels (typical of the toughest permits) even though these machines have five times the turbine-exit NOx levels.” Plus, they likely will be required to cycle and operate at less than design output for significant operating hours over their lifetimes.

Fig 5 indicates that for these performance specifications, you’ll either have to accept high operating and compliance risk at the ragged edge of the capabilities of traditional single-bed SCRs or resort to a more complicated and more expensive SCR design.

When you operate advanced-technology machines at low loads, you tap out the capabilities of the design (Fig 6). “The ammonia injection grid can’t handle both the NOx levels at the maximum design output and what would be typical at 30-50% load, because of the corresponding changes in mass flow, temperature, and mixing.”

Environex specialists believe owner/operators of G- and H-class machines will have problems because vendors are supplying SCRs with inadequate catalyst volumes. It’s also important to consider adding a permanent grid made of stainless-steel tubing within the HRSG housing which allows you to periodically take 2-D distribution-grid measurements for NOx and NH3 and more accurately tune the AIG distribution valves.

Such capability nominally adds about $50,000 to the budget, a sum that looks paltry compared to the penalties of non-compliance.  The high NOx conversion requirements for these systems coupled with low ammonia-slip limits decrease the tolerance for non-ideal ammonia-to-NOx distribution. You should expect this to increase the required frequency for ammonia-grid tuning. 

Duct burners, of course, are another source of NOx and CO which must be accounted for through real operating data. SCR inlet NOx can more than double at design GT output and full duct firing and the SCR operating temperature can climb by 50 to 100 deg F. Inlet CO, meanwhile can decrease. During interim periods as duct burners come up to full capacity, or remain at part load, NOx and CO emissions can be quite variable. These impacts can be quite striking if your unit was designed for baseload operation.

Generally, Toback concludes, increases in exhaust flow to the SCR impairs catalyst life, while decreases in SCR inlet NOx and CO emissions and increasing SCR operating temperature extend it. If your machines are no longer operating the way the SCR was designed, it’s time to consider a program to acquire the operating data needed for optimizing the SCR process for new conditions.

Posted in CCUG | Comments Off on Consider the impact of new operating regimes on your SCR

Steam drums demand regular inspection, continual attention

Ever stop to think about what’s going on with your HRSG’s steam drums, out of sight for the most part and all wrapped up in insulation? Even if you were to hike up to the drum level what could you see?

Next question: How many people at your plant really think about the boiler? Their primary focus is at ground level, on the gas turbine/generators. The O&M team knows a great deal about the GTs given inspections every six months or so and the diagnostic instrumentation typically installed. Lifecycle data are tracked continuously.

How are you tracking the condition of your HRSG and its remaining life? Anything much more than looking for steam/water leaks and monitoring drum level and steam pressure/temperature, and an occasional inspection? At many plants, probably not.

That’s why the Combined Cycle Users Group (CCUG) steering committee invited HRST Inc’s Lester Stanley to talk about steam drums and their care at the organization’s 2017 conference in Phoenix at the end of August. He has been inspecting boilers of all types his entire professional career and likely crawls through more drums in a given year than the overwhelming majority of users do in a lifetime.

These thick-walled steel behemoths may seem indestructible, but they are very susceptible to life-threatening cracking and other problems when not properly operated and maintained. Stanley encouraged users to pay close attention to OEM-recommended warmup and cooldown procedures, closely monitor ramp rates, conduct regular inspections, and pay prompt attention to repairs suggested by experts.

He began his presentation, available to users online, with a quick primer on steam drums, their function, and refresher illustrations, and a list of five important drum components (Fig 1). Then he identified the five issues he would speak to:

      • Steam-purity degradation.

      • Drum-level-measurement piping condition.

      • Nozzle weld cracking.

      • Shell weld cracking.

      • Manway sealing reliability.

“Steam-drum performance and steam purity were probably carefully tested at commissioning,” Stanley said. But in the second half of life, “age is affecting the mechanical condition of the drum internals.” This can result in excessive water carryover. Specifically:

      • Primary and secondary separators get fouled with rust.

      • Fouling occurs in the final separators, increasing velocity and impairing moisture separation.

      • Gaps appear in the final separators, allowing wet steam to pass through.

      • Separator housing and supports fail from stress and also allow wet-steam bypass.

      • Errors occur in drum-level sensing, leading to higher water levels and less volumetric space for moisture to drop out; this is usually caused by level-transmitter calibration/compensation.

Detecting these problems is best done through more frequent saturated steam sampling, and even continuous on-line monitoring.

The condition of drum-level piping is critical for accurate level sensing. Age-related risks include plugging of sensing lines with debris and corrosion of the small-diameter piping. The latter risk grows with increased downtime. Removable insulation blankets exacerbate rainwater ingress. Solutions are to inspect under insulation more frequently and consider use of removable insulation with better protection against rainwater.

The focus for nozzle and shell weld cracking is the high-pressure (HP) drum but all steam drums should be inspected, Stanley urged. Although the intermediate-pressure (IP) drum is the least susceptible to weld cracks, it’s easy to include it in the inspection program for the HP and LP so you “sleep better at night.”

On/off cycling drives weld cracks in the HP drum. This causes temperature differentials between the shell and its nozzles because the drum pressure cycles across a wide range, such as from 0 to 400 psig. Problem is, there are numerous thick nozzles and shell welds in an HP drum. In the LP drum, internal pitting, leading to cracks, is the notable threat.

Stanley then spent considerable time reviewing several relevant inspection techniques and offered suggestions for how to prepare for and set up an inspection program. Perhaps the overriding message was, don’t try this at home; in other words, retain an experienced, qualified crew.

Manway reliability was Stanley’s final topic. The ageing risk is the deterioration of the gasket sealing surface over time from cleaning and removing old gasket material, impurities in water residue, gouging from steam cutting or closing damage, and corrosion during downtime. The gasket sealing surface has to be smooth but not “polished” smooth, and there’s more precision necessary in the serrations (like an album surface) than you might think.

There are a few specialists with the proper tools to re-machine the sealing surface, but sometimes replacing the door may be the best option. This allows you to ease the problems associated with achieving the minimum sealing stress for various types of gaskets. The proper seal stress is difficult to achieve with studs and a torque wrench because drum pressure in operation provides the last increment of stress necessary. Many HRSG OEM manuals require hot retorquing procedures, which are dangerous and should be avoided.

Stanley thinks a manway door with Belleville washers is a better design (Fig 2). Eliminating hot retorquing allows you to install a steam shield for additional personnel protection. Stanley noted that the Belleville washer design has gone through several years of demonstration in the field.

End notes. Encouragement for showing your steam drums greater respect might come from within if you think about the difficulty associated with making Code-qualified repairs to the thick-walled vessels and, in the extreme, replacing a drum. Fig 3, provided by Bremco Inc, offers some perspective.

Development of diagnostic tools and inspection techniques for tracking the life consumption of steam drums is ongoing. Someone at your plant should have the unofficial title of HRSG King (or Queen) with the responsibility for attending one meeting annually to keep up with these developments and the experiences of other owner/operators in their implementation.

If you don’t have a person on staff with deep HRSG experience, you might consider selecting someone from your O&M team to attend HRST’s rigorous three-day HRSG Academy (next session will be in Tucson, Ariz, Jan 23 – 25, 2018) to get the foundation necessary to help guide the plant’s boiler decision-making.

With a background in fundamentals, annual attendance at the HRSG Forum with Bob Anderson is recommended by the editors to keep up with technology, methods, procedures, equipment, services, etc, of importance.

Finally, it’s vitally important to incorporate the latest experience into your specifications for new HRSGs. Electronic cutting and pasting of a years-old spec will restrict your capabilities in such things as fast starting, fast ramping, etc, and possibly leave you vulnerable to current issues—such as poor control of attemperators and steam bypass systems.

Modern control systems and upgraded materials enable reduced drum storage volume, smaller drums, and thinner shells. Use of stack dampers, high-quality valves, steam sparging, robust modern instrumentation, etc, can maintain drum metal temperature above the recommended minimum during shutdown periods to maximize life.

Posted in CCUG | Comments Off on Steam drums demand regular inspection, continual attention

HRSG best practices: Plan now for old age

It happens to all of us. One day we wake up and realize, we’re not young anymore. You can, in the same breath, decide you need to do some things differently. Or, you can ignore it, and pay a larger price later.

The same thing is true with heat-recovery steam generators (HRSGs). Combined-cycle plants experienced the equivalent of the post WWII “baby boom” between 2000 and 2005 (Fig 1). These units are now in or approaching the second half of their 25-30-year design life.

The need to acknowledge this reality and plan accordingly was considered important enough that the Combined Cycle Users Group (CCUG), at its 2017 conference in Phoenix at the end of August, devoted the better part of a morning to the subject so that experts from HRST Inc could identify some the problems users could experience and possible solutions.

At an industry level, Bryan Craig noted that the general goal is to plan for and avoid similar ageing problems experienced with the fleet of fossil-fired boilers installed a generation earlier. The range of issues presented, and the photographic evidence from operating units, in the HRST slides is so extensive that users are strongly encouraged to access the original presentations. This article gives some highlights.

Like many fossil units, combined cycles often operate in ways not designed for—that is, less baseload and more cycling and dispatch. Some of the original design materials and methods may be questionable as well.

Creep and overheat damage in superheater and reheater tubes is the first problem Craig tackled. Most tube overheat incidences occur downstream of the duct burner and are caused by flame impingement. Flames should never make contact with tube metal yet they often do. Rules-of-thumb for flames are that they should be 6-10 ft long, they should be independent and separated, and reach one-half to two-thirds down the firing duct.

While users should view flames at least once daily, and preferably once per shift, through the unit’s viewports, a better idea is to install cameras in the firing duct on the walls, floor, and/or ceiling and wirelessly transmit the images to screens in the control room. Damage is often worse in areas difficult to view through the casing ports.

Up next were duct-burner problems, notably baffle sagging, cracking, and fluttering; flame-holder failures; coking; burner-nozzle cracking; and flow-distribution equipment failure. In the case of baffles, for example, they often sag under their own weight as the radiant heat weakens the metal. Because they also provide horizontal and vertical support to the burner elements, weakness in the baffles causes problems in the burners, such as fatigue cracks. Fig 2 illustrates the some of the issues with one type of burner and vintage, and the fix HRST has implemented.

Casings experience cracking and corrosion problems at the roof because of numerous piping penetrations, and on the sides where temperatures generally above 800F may exist. Unrepaired casing cracks can lead to graphitization and more extensive repair work. Of course, any cracks, especially in the roof, lead to rainwater ingress and insulation failures, which only compound problems with internal surface corrosion. 

In the duct-firing area, there’s more insulation to protect against higher temperatures. However, when the burners are not running, this becomes the coldest area of the casing. If the temperature dips below the exhaust gas dewpoint, acidic condensation may occur, leading to rare cases of stress-corrosion cracking. High NOx environments, units with no SCR for example, are especially prone to SCC.

Ageing problems in economizers. Craig focused on the return-bend style of economizer design (different from panelized economizers). Much of the economizer’s weight is supported by the return bends, making them prone to corrosion fatigue. Startup thermal shock aggravates the situation. To address this, HRST has developed a retrofit support system that avoids any new pressure parts.

Craig offered suggestions and more robust inspection, testing, and assessments for addressing high-pressure evaporator waterside deposits and high-temperature piping. The API 579/ASME FFS-1 specification to assess “fitness for service” and remaining life should be considered for high-temperature piping. He noted that a “high percentage of welds being tested are problems.” For areas prone to corrosion under insulation, Craig suggested retrofitting vents and drains.

Posted in CCUG | Comments Off on HRSG best practices: Plan now for old age

RO Part V: When to clean, how to clean

This is the final part of a five-part series on the design, operation, and maintenance of reverse osmosis systems for powerplants compiled by Wes Byrne, U.S. Water’s consultant on membrane technologies. Parts I, II, III, and IV are identified below.

      • Part I: Value proposition, how it works

      • Part II: Importance of a pilot study in system design

      • Part III: Mitigating scale formation and membrane fouling

      • Part IV: System operation and monitoring 

Chemical cleaning is a routine requirement for most RO systems. Frequency depends on the effectiveness of the pretreatment equipment.

As fouling solids or scale particles accumulate, their characteristics often change and they become more resistant to cleaning. Clay and biological materials will tend to compress against the membrane surface and become chemically resistant as water is squeezed out of their structure. Scale formations may change from being primarily calcium carbonate (relatively easy to clean) to calcium sulfate (difficult to clean).

The change in normalized RO performance variables can be used to determine cleaning needs. Most membrane manufacturers recommend cleaning before these variables change by about 15%.

Certain types of fouling solids or scaling salts may have a substantial impact on permeate quality. Aluminum salts may come out of suspension as a fouling particle, only to re-dissolve if the water acidity changes. This may then result in increased aluminum passage from the membrane surface through the membrane and into the permeate/purified water.

Calcium carbonate scale may leach a relatively high concentration of calcium carbonate through the membrane into the permeate stream and affect the conductivity. Most other fouling solids will not have a significant impact on RO salt rejection unless the fouling is extreme.

Membrane cleaning involves passing a cleaning solution through the membrane system at conditions that promote the dissolution or delamination of the fouling solids from the membrane surface or from the spacing material along the membrane flow channels. The optimum solution will depend on the particular fouling solids or scale particles, and the relative ability to clean will often be limited by membrane chemical tolerance.

Most strong oxidizing agents that would typically be effective in cleaning biological solids are not going to be compatible with the RO membrane. There will also be limits to the pH extremes that should be used. In addition, while higher temperature will increase the rate of cleaning, the solution temperature will be limited to below 105F or as designated by the membrane manufacturer.

The most critical characteristic of a cleaning solution is its pH. Acidic solutions are more effective in dissolving metals and scale formations, while alkaline (high pH) solutions are more effective in removing clay, silt, biological, and other organic solids. Strongly acidic solutions may stabilize biological solids and therefore should not be used as a first cleaning step. Finishing a cleaning with a strongly acidic solution will tend to leave the membrane with increased rejection characteristics but somewhat reduced permeate flow, while finishing with a strongly alkaline solution will have the opposite effect.

The addition of specific cleaning agents often improves the solution’s cleaning abilities. A chelating agent assists in pulling out metals from the fouling solids, while surfactants/detergents improve the solution’s ability to penetrate the fouling solids and suspend oily substances. The use of surfactants may reduce cleaning time but will increase the time required for rinse up.

When the fouling solids are causing a flow restriction, increasing normalized pressure drop, high cleaning flow rates (within the membrane manufacturer’s guidelines) through the membrane feed channels will cause agitation that will assist in breaking up the deposits. When the solids coat the membrane surface and reduce the normalized permeate flow rate, the delamination of these solids will be most easily achieved if water is not permeating through the membrane during the cleaning process and creating a force that holds the solids to the surface. This means cleaning at low pressure.

Achieving a high cleaning flow rate that is balanced throughout all of the membrane vessels usually requires that each vessel stage be cleaned separately. This also helps minimize the pressure required to push the solution through the elements. Cleaning solution is therefore pumped at high flow rates, as recommended by the membrane manufacturer. It is pumped at the maximum pressure required to achieve the target flow rate, but may be limited to 60 psi to reduce the potential for crushing or otherwise damaging the membrane elements.

The solution is directed in the normal feed-end direction of flow and the exiting concentrate stream is then returned to the cleaning tank at minimal backpressure. The flow direction may occasionally be reversed so that the solution enters the concentrate end of the stage when fouling solids are blinding the face of the lead-end membrane elements.

There may be a small flow of permeate that should also be returned to the cleaning tank using a separate line. In spite of its low apparent flow rate, the permeate should never be valved off because this may put certain membrane elements at risk of physical damage.

Data should be recorded during the cleaning process. With membrane surface fouling, it is difficult to gauge when original performance has been restored until the unit is rinsed and operated normally. If the fouling solids were causing an increased pressure drop in the RO, then the cleaning inlet pressure can be used as a measure of cleaning progress. If the pressure keeps declining, the cleaning is still removing fouling solids. If the fouling is severe, it may require a number of hours of circulation before the inlet pressure stabilizes.

Cleaning success is confirmed when the normalized pressure drop and normalized permeate flow rate return to their startup values.

Posted in CCUG | Comments Off on RO Part V: When to clean, how to clean