Changing the way the Power Generation Industry buys parts.

Detailed presentations on DLN-1 troubleshooting, ZLD solutions, varnish prevention identify outage actions to improve plant performance

The 2007 meeting of the 7EA Users Group, October 9-11, in San Francisco, was the largest in recent memory with 110 owner/operators attending and 78 products/services suppliers participating in the vendor fair on Wednesday evening. The technical program developed by the steering committee (Sidebar 1) was top-notch.

Logistics for the conference and trade show were managed by Greg Carvalho of Simplified Technology Co, Fremont, Calif; he will coordinate the 2008 meeting as well. Carvalho is probably best known in the industry as the host-master for e-mail forums available at no cost to more than a dozen user groups, including the 7EA.

This report summarizes prepared presentations by both users and other industry experts that offer valuable input for both planning and conducting maintenance outages. The user presentations come first. They discuss how the use of simulators in operator training can help reduce unit trips and extend maintenance intervals, the value of group participation in the development of O&M best practices, and compressor and turbine issues identified during a recent 7EA outage.

Non-user presentations included the following: Areas of zero-liquiddischarge systems to investigate during outages to correct operating problems, how to troubleshoot DLN-1 combustion systems, and the various alternatives available to mitigate varnish formation on critical components in lube-oil and hydraulic systems.

User presentations

Dave Ulozas, the new chairman of the steering committee, opened the technical portion of the meeting with a presentation on the use of simulators in operator training programs. He is unique among plant managers. Ulozas is responsible for a 2 × 1 7EA-powered combined cycle as well as for a coalfired station that shares its site with the grave of an early nuclear reactor. Nebraska Public Power District (NPPD) is his employer.

A backgrounder on the plants before moving ahead: Beatrice Power Station, 250-MW combined cycle named after its “home town,” began commercial operation in January 2005 and has operated in intermediate- load service since. Site was selected based on its access to natural- gas pipelines, existing electric transmission infrastructure, and water supply. Plant is operated by a permanent staff of 14.

Sheldon Station consists of two coal-fired steam units with an aggregate capacity of 225 MW. Both burn Powder River Basin coal today. The first unit began operating in mid 1961, the second in mid 1968. Total staff numbers 95.

Hallam, the nearest town, was the name adopted for the 240-MW, sodium-cooled graphite-moderated nuclear powerplant built on the Sheldon site at the same time as the first coal unit. The Atomic Energy Commission (since absorbed into DOE) dismantled Hallam between 1967 and 1969 after it satisfied operational objectives. Some components were buried onsite, the rest shipped elsewhere.

A goal of every facility manager is to maintain plant staff at full complement with capable people in every position. Not much different than baseball when you think about it. Managers of gas-turbine-based generating assets typically supervise young staffs; their counterparts at coal-fired stations, much older ones.

The personnel risk associated with the former is restlessness and the temptation to jump after a marginally higher paycheck; with the latter it is retirement. In both cases, new people have to be hired and trained, and more likely than not your replacements will be inexperienced in powerplant operations—especially if facilities are located in the middle of Nebraska like Ulozas’ are.

Plant managers can waste no time getting a new hire up to speed. If the person requires basic indoctrination in powerplant equipment, systems, and processes, perhaps Internetbased coursework like that available from General Physics Corp’s GPiLearn™ (www.gpworldwide.com/ energywbt) is a good place to start.

To teach someone his or her job, and to make that person a productive member of the plant staff, there’s probably nothing more effective than teaming up the new hire with an experienced employee. However, GTbased plants have so few employees that teaming has practical limits.

In coal-fired plants, O&M knowledge resides in the minds of employees who may have one foot out the door. Where’s their incentive for training a new hire? Professional pride, perhaps, but there can be a shortage of that at any given plant.

Ulozas is a champion of quality training. “It helps to make powerplant operations more predictable,” he said. “You want procedures, such as startups and shutdowns, to be repeatable; no mistakes.” Simulators are a big help in this regard, he continued.They require that the trainee make decisions in real time based on multiple inputs. In so doing, simulators improve their knowledge and develop the necessary operational skills.

Most people agree that one learns from his or her mistakes; also that it is far better to learn from mistakes made on a simulator than in the control room of an operating plant.

Ulozas said a quality simulator is a valuable tool beyond practicing startups and shutdowns and how to respond to abnormal events and accidents. It can be used to evaluate options for improving heat rate, teach teamwork, train personnel on the proper procedures for running equipment/system diagnostics, assist in the development and validation of new operating procedures, etc.

For those unfamiliar with simulators, Ulozas made it a point to differentiate between emulation and stimulation. In the former, the DCS (distributed control system) software is replicated within the simulator code; in the latter, actual DCS software is used to interface with the simulator code.

The difference between these alternatives is important, he stressed. To illustrate: The stimulation option permits the training program for E&I technicians to practice controller tuning on the simulator; for the engineering staff, it allows testing of proposed plant modifications prior to implementing changes. In addition, it is useful for systems and integrated- plant training.

Ulozas suggested that anyone considering the purchase of a simulator should review EPRI (Electric Power Research Institute, Palo Alto, Calif) report AD-103790, “Simulator procurement guidelines for fossil powerplants.” It contains a detailed procurement specification, including payment schedule and methodology.

What’s particularly important for ensuring that your simulator is of the highest quality and greatest benefit to your staff is to make actual plant data available to developers. If you manage an old plant, like Sheldon Station, consider running detailed performance tests to provide accurate information.

Ulozas said simulator cost varies with its fidelity. A low-fidelity “suitcase simulator” that’s acceptable for training operators on startups and shutdowns can be had for a few hundred thousand dollars. Medium fidelity, which allows you to change variables, can cost upward of a million dollars. A high-fidelity simulator— the one you really want—can run a million and a half, or more. However, it’s tough to get that degree of corporate financial support.

Don’t forget, Ulozas reminded, that instructors must be trained on how to use the simulator for training plant personnel, how the models work, and what assumptions were used in their development, etc.

Next, he presented a case history on the development of a simulator for Sheldon Station and implementation of an effective training program. The first attempt, Ulozas confided, fell short and offers good experience for others. Thorough planning, commitment, and attention to detail are critical to the success of your project, he continued.

First work on an emulator for Sheldon began in 2000. Plant data (drawings, design-basis information, test results, etc) were generally poor, resulting in the design of a substandard simulator. Its capability essentially was limited to startups and operators had a low opinion of the model’s validity.

Ulozas and his supervisors went back to the drawing board knowing full-well that some 30-year veterans were planning their retirement parties. Modifications were made to the simulator, fidelity improved, and its use in a formal training program commenced. First goal was to have each watch crew develop more completely as team. It required practice in diagnostics, communications, and conservative decision-making.

Each crew member keeps a notebook to record improvement areas and lessons learned during training, which is conducted during the watch. Design of logical scenarios for personnel development is important to achieving objectives. One of the challenges facing supervisors is to identify training opportunities within the scenario. For example, if the scenario includes transfer of the voltage regulator to manual while a load change is in progress, it offers the opportunity to review the theory voltage regulation.

It also offers the opportunity for system review—that is, a review of drawings to determine what will cause the voltage regulator to transfer to manual. Team performance can be evaluated, too, by reviewing how well the crew communicated and practiced peer-checking during the event.

A scenario begins when the instructor gives the crew a turnover sheet with initial plant conditions and the crew announces that they have the watch. Important to note is that a scenario can be put on hold by any crew member, or the instructor, to clarify points, to explain something, etc. When the scenario is completed, the instructor summarizes what was achieved and the crew critiques itself using a worksheet provided. The worksheet is comprehensive.

For example, one of the competencies that must be evaluated is control- board operation. Here are some of points requiring evaluation:

• Ability to locate the controls required to handle the event.
• Manipulate the controls accurately and in timely fashion.
• Act appropriately in response to instruments.
• Take manual control when appropriate.
• Demonstrate self and peer checks.

Other competencies included on the training critique worksheet are these:

• Diagnosis of events.
• Understand plant/system response.
• Adherence to and proper use of procedures.
• Communications.
• Understand and interpret alarms.
• Direct shift operations.

Ulozas is the first to admit that learning how to use the simulator as a training tool is half the battle. Since that was achieved at Sheldon, tremendous benefit has been realized.

Users working collaboratively

Pat Myers, plant manager of American Electric Power Co’s Ceredo Generating Station in Huntington, WV, is a frequent speaker at 7EA User Group meetings and known among peers as a person willing to share his considerable knowledge. His presentation at the San Francisco meeting was aimed at getting others to do the same by participating in a formal program of experience-sharing and preventive-maintenance (PM) planning that he volunteered to lead.

Myers opened with a question: What should you be doing regarding PMs? His observation: There’s not enough staff at any gas-turbine based plant to comply with all the manufacturers’ recommendations. That’s a problem for some of the less experienced supervisors, he said.Overwhelmed by the challenge of having to do everything, they tend to do little more than focus on management and regulatory directives.

But somewhere between “nothing” and “we have to do all this” is the answer, Myers continued. Having best-practices guidelines for preventive maintenance would help virtually everyone in the group improve his or her plant’s performance and probably reduce expenses as well.

He spent a considerable amount of time compiling a presentation that identified about a dozen issues/tasks tackled by plant supervisory personnel that might be resolved/facilitated by developing best practices based on 7EA user experience. Myers started with lube-oil filtration, which when done correctly can help mitigate varnish formation.

One of the first things the group should investigate, he suggested, is why some units are more prone to varnish formation than others. Some of the reasons would come later in the program during a panel discussion on the subject. Myers said his experience indicated that tighter control of the main lube-oil filtration package might be all that’s needed.

In 2005, Ceredo’s main filters were upgraded to ones with a beta-ratio rating that would ensure removal of 99.5% of all particles 3 microns and larger. Lube-oil health with the new filters compared favorably with that for two units onsite equipped with additional kidney-loop polishing filters that ran continuously for six months.

Other information and findings important to this discussion: Plant has used Chevron GST 32 since commissioning in 2001. Units average approximately 250 starts and 1000 operating hours annually; they also average about eight hours weekly on the ratchet. Varnish potential rating is well below 10 and ISO 16/14/11 is the cleanliness standard maintained.Findings: no varnish buildup in lastchance filters; no control problems related to varnish deposition. The question: If the Ceredo 7EA plant is having excellent success why are other plants having trouble?

Regarding the ISO spec, it is the suggested hydraulic fluid standard to help prevent fouling of servo valves. Translation: The 16 means the permissible number of particles 4 microns and larger in a lube-oil sample of 1 milliliter (ml); 14, particles 6 microns and larger; 11, particles 14 microns and larger, as indicated on the ISO Code table.

Myers recommended several other issues/tasks that would benefit from the development of best practices, including these:

• Battery capacity testing, because of the problems that Ceredo observed in its first five-year test. In response to the issues identified, the plant assembled emergency battery “jumper” cables, and an emergency battery disconnect for use in situations where batteries or chargers may fail.
• Dc lube-oil pump testing and operation. Myers strongly recommended annual testing. His presentation mentioned the value of TIL (technical information letter) 1469-2R1, which addresses annual inspection. He also discussed his plant’s modifications of pump operating logic to incorporate an automatic dc-pump test at each start, and the steps to incorporate the latest updates recommended by the OEM (such as those in TIL 1420-2). Further, that fleet experience be shared to be sure any anomalous operating condition which might compromise proper dc lube-oil pump operation be identified so others could make logic changes to prevent its occurrence.
• Compressor bleed-valve trips. Ceredo’s experience is one trip caused by a bleed valve not closing in time during a start in extremely cold weather and six trips caused by bleed valves not opening in time.

Myers and his staff believe that the root cause of all the trips attributed to “valves not opening in time” was improper operation of solenoid valves caused by use of the wrong spring material in the valves. He showed pictures of “dissected” valves to illustrate the wear and tear of the all-important solenoid spring and noted findings (from another user) of bleed valve cylinder sealing “O”-ring distress, Viton seal-ring failure, bearing failure on turning yoke, and shaft bushing displacement.

Myers asked the group: What is the proper time to replace hardware/ consumables? What are the best techniques for repair/replacement? Has anyone installed a system to trend valve motion versus solenoid energize/de-energize?

Obviously, his point was that everyone was dealing with the same challenges and the wheel was being reinvented in each plant, and for what reason? The 7EA users certainly could come up with a best practice for solenoid testing and maintenance that would improve fleet starting reliability.

• Torque-converter spool setting to ensure proper purge cycle. Myers discussed a problem that he encountered on a Voith hydraulic drive that caused increasingly slower unit start times. What he found was a loose stop screw that caused the spool to have less and less travel on successive starts.
• Ratchet diagnostics. Myers asked the group, “Does anyone have any kind of automatic shaft-movement verification and alarm when the unit is ‘on ratchet’ [turning gear] but the shaft is not moving? He then discussed his plant’s experience with the old-style hydraulicratchet pressure switches, which are prone to diaphragm failure, and the new-style, which are much more reliable.

Myers also mentioned logic changes to Ceredo’s standard Mark V control system to improve ratchet operational reliability. You knew from just looking around the room that everyone had not made all these logic changes and that a 7EA Best Practices initiative would be beneficial.

Some very ugly photos of oil spills caused by fittings breaking on the hydraulic ratchet were shared. Myers reported after his presentation what he had learned from a Voith representative at the vendor fair: A setting on the control of “7” or less and reinforcement of the piping should help. Myers will share a piping-support design he has underway with the 7EA users via the group’s message board.

• Secondary flame scanners. He gave a brief history of how the secondary flame scanner’s view tube was significantly shortened to increase flame counts.
• Testing of the fire protection system. A detailed listing of Chemtron checks and a recent inspection report were shared, as was a relay wiring modification implemented to improve reliability of the ice cube relays on the Chemtron panel.
• IGV (inlet guide vane) block to ensure personnel safety with lock out/tag out in effect. Myers shared a locking-block design that was implemented at Ceredo for a reliable LO/TO when working through IGVs during borescope and other inspections.
• Icing detection. He proposed a new design for the bellmouth door to assure positive visual identification of icing. The door design is much lighter than the original duct door and provides visual access to detect water in the duct and to observe IGV icing.
• Turbine trips and run-backs. “How ‘deep’ do you go in verifying proper operation of turbine trip devices? How ‘deep’ should you go?” After listening to Myers for a couple of minutes, you realized that this might well be another area that would benefit from a best practice. He provided a detailed listing of trips and runbacks that can be used to design a plant-specific turbine trip check protocol.
• Overspeed testing. Next, he shared a proposed procedure for electrically testing overspeed test circuits without actually stressing the rotor with an actual overspeed event
• Transferring to premix steady state during a start in very cold weather Myers finds can be problematic. He described a problem he is having with one of his units and believes that collaboration among users could help everyone—even him.
• To close his presentation, Myers shared a “lone worker” alarm system he has implemented at his plant to add an additional level of safety communications for employees required to be on site and work “solo.”

Other user presentations

Olaf Barth of Dominion Energy Inc’s CT Operations unit reviewed his company’s spring outage experience, focusing on areas of concern in both the compressor and turbine sections of its 7EAs. Paul Beatty, O&M superintendent for Duke Energy’s Lincoln, Mill Creek, and Buzzard Roost peaking stations, updated the group on test runs of one 7EA on 100% biofuel.

Environmental stewardship

A virtue of gas-turbine-based plants is that they require much less water for operation than fossil-fired and nuclear steam/ electric plants. This is particularly important today because water consumption has become a high-profile discussion topic at siting hearings nationwide. Constrained potable water supplies in “unlikely” places, such as Atlanta, are a reason for growing public concern.

Use of so-called “grey water” now is mandated for powerplants in many areas, by law or by just good business sense. Plus zero-liquid-discharge (ZLD) and near-zero-liquid-discharge systems have become necessary addons where owners want their new generating plants licensed in timely fashion. These systems protect the environment against potentially harmful liquid discharges; however, their real reason for being, in most cases, is to wring out and recycle every last drop of water from the powerplant waste stream.

Most owner/operators are not familiar with today’s wastewater treatment systems, but they should get to know more about them sooner rather than later. These systems deserve your greatest respect. They can be extremely complex and pose O&M challenges not encountered in the electric power industry since the introduction of SO2 scrubbers on coal-fired stations in the early 1970s (Fig 1).

Among the 7EA users, perhaps no one understands this better than George Davies, combustion turbine department manager for Turlock (Calif) Irrigation District’s Walnut Energy Center, and former member of the group’s steering committee. Walnut is a 250-MW, 7EA-powered 2 × 1 combined cycle with a “stateof- the-art” ZLD system. It also is the only powerplant known to the editors that has outsourced ZLD operation and maintenance to a third-party services firm (CH2M Hi l l subsidiary Operations Management International, known as OMI).

Credit the steering committee for its forward thinking in recognizing the future information needs of the organization’s members and for inviting Dan Sampson, Nalco Co’s (Naperville, Ill) power-industry technical consultant, to speak about the challenges presented by ZLD. Sampson, one of relatively few experts on the subject, has been involved handson in powerplant wastewater treatment for well over a decade. Before joining Nalco, he was with Calpine Corp, San Jose, and had direct involvement in system installation, commissioning, and redesign to correct deficiencies.

GT users have to bring their “A-game” listening skills to a Sampson presentation. Most have had little exposure to water-treatment technology and terminology, and ZLD certainly is a step beyond the norm. Blink and you can be lost for the rest of the presentation: That’s how complex the practical application of this technology is. It’s fair to say that if this was an attendee’s first exposure to ZLD and he or she wasn’t intimidated the user has to be an extraordinary individual.

Perhaps the best place to begin is with Sampson’s concluding remarks.“You are not alone,” he said, “just about everybody hates the ZLD system they have.” All ZLD systems have problems, Sampson continued, vendor designs haven’t improved much over time. Problems identified a decade ago are still evident in new installations.

To stay current and to get the practical advice needed to ensure continuity of operations, he suggested active participation in user groups that address ZLD issues. Keep in mind that ZLD systems are “must run” facilities, just like SCRs (selective catalytic reduction) for NOx control. If those systems do not operate, in all probability, neither can the generating plant.

Sampson’s 45-min presentation was one of the most comprehensive you’d hear at any user group’s meeting. His format was problems/ solutions in the following subject categories: general, clarifier/softeners and filters, ion exchange, wastewater reverse osmosis (RO), and thermal systems (brine concentrator and crystallizer).

To give you an idea as to the depth of coverage, here’s what Sampson had to say about membrane fouling, which many users have experienced (Fig 2). The first segment of the twopart presentation covered fouling typically encountered. He said it usually results from calcium sulfate and carbonate scales, ineffective removal of suspended solids and iron ahead of the RO unit, and silica.

These problems generally are mitigated by changes to the operation of both upstream equipment and chemistry. For example, do a more effective job in removing calcium through better operation of the cold-lime softener, sodium zeolite softener, and hydrogen-cycle WAC (weak-acid cation exchanger, Fig 3). If alkalinity is problematic, check both the decarbonator and the acid feed to the decarbonator for proper operation.

Improving the removal of suspended solids and iron demands proper ferric coagulant feed, anionic polymer feed, and TSS (total suspended solids)/SDI (silt density index) monitoring (Fig 4). Stabilization of silica is assured by upstream removal or high-pH operation. Other recommendations include:

• Monitor normalized data.
• Keep a spare set of membranes onsite.
• Perform frequent membrane autopsies.
• Provide a wash skid that allows you to clean membranes in place.

High pH. For ZLD systems operating with an RO feed pH above 10, even a small amount of hardness causes membrane scaling. To combat the formation of a debilitating scale, Sampson recommends periodically reducing the water recovery of the RO train to 60% to 75% of the design operating value, or less, and also injecting some acid into the feed line.

Oftentimes, just stopping the flow of caustic used to raise the pH of water entering the RO unit is sufficient. Reason is that the decarbonator, located immediately ahead of the RO unit, typically discharges water at a pH low enough to meet requirements. Run the RO unit for a short period at the lower pH to dissolve carbonate scale. The calcium sulfate scale formed during this step is dissolved when you return the system to high-pH operation. Final step is to ensure the recommended flow of anti-scalant.

Sampson stressed that manpower is at the root-cause of many ZLD problems. The most common mistake, he said, is assuming that ZLD systems can be operated with the same staff as a typical water plant. Not so. ZLD systems are much more complex and the manpower requirement varies with complexity.

“Simple” ZLD systems alone require from seven to nine experienced and dedicated personnel.Positions include one operator 24/7, half-time mechanic, half-time I&C technician, and a full-time manager (either a plant chemist or O&M manager).“Complex” designs require 12-14 people beyond the normal powerplant complement, specifically: two operators 24/7, full-time mechanic, full-time I&C technician, and manager.

“No ZLD system runs itself,” Sampson barked. “Vendor sales pitches have created the wrong impression.”You need at least one experienced person constantly monitoring operating parameters and running chemistry— no other duties—plus at least one other person available to respond immediately to operating problems and correct them. He noted that it takes only a matter of minutes to foul up system chemistry to the point that it can take days to correct.

With many attendees looking ahead to their first experiences with ZLD, Sampson offered a few rules-ofthumb in case these users are in a position to influence design decisions.

Developers, he said, are prone to buy the lowest-price system designed to handle the required flow. One problem with this approach is that the reference water analysis for design work usually is uncertain and any deviation from design almost always translates to a loss in capacity. Consequently, ZLD systems almost never meet their nameplate ratings.

For decision-making, assume the following:

• Mechanical reliability, 75% to 95%; assume a nominal 80%.
• A 20% degradation in output between system overhauls/cleanings.

This translates to an effective capacity for the ZLD system of 64% (0.8 × 0.8). In simple terms, if you need a system to handle a 300-gpm waste stream, specify 500 gpm. This is the best case for effective capacity; more complex designs have more things to go wrong and the design margin must be greater.

Recommended sizing criteria based on maximum peak flow:

• “Simple” designs (those relying only on a brine concentrator), recommend two 60% trains.
• “Complex” designs (those using RO to concentrate the wastewater stream before it flows to a crystallizer or other downstream equipment, Figs 5, 6), recommend two 100% trains.

Risk assessment must be part of the initial plant design effort, continued Sampson, and your cost analysis should include redundancy. Don’t forget to perform a single-point failure analysis and to purchase additional shelf or installed spares (pumps, filters, etc) regardless of system design.

DLN-1 troubleshooting

Mitch Cohen, a senior systems engineer for Orlandobased Turbine Technology Services Corp, is respected by many readers for his knowledge of large frame fuel systems (visit www.combinedcyclejournal.com/archives.html, click 3Q/2005, click “Improve GT operating flexibility, reliability with fuel-system mods.” At the 7EA User Group’s San Francisco meeting he conducted an hour-long clinic on troubleshooting DLN-1 operating problems—a specialty of his—that was well-received by attendees.

Cohen spends a great deal of time in the field so he knows well the combustion- related issues that confront plant personnel during “normal” operation, plus those that torment plant managers as they try to bring their units into air-permit compliance after a maintenance outage.

Think of him as a country doctor for GTs. During the presentation, Cohen asked on several occasions, “Has anyone experienced this problem or one like it?” A couple of times, users described incidents of concern and he offered diagnoses and “next steps” to resolve them. Owner/operators who don’t participate in the user group for their engine model miss out on opportunities such as this to access the expertise of the industry’s top solutions providers.

Cohen began his presentation with three schematics for the 7EA DLN-1 to get attendees “on the same page”: combustor (Fig 7), fuel system original configuration (Fig 8), and the new configuration for the fuel system (inset gas skid with Fig 8). Most users had units outfitted with the original fuel-system configuration. It has one gas control valve to serve the primary, secondary, and transfer fuel streams, plus two three-way splitter valves. The new configuration features three gas control valves arranged in parallel, each dedicated to a given fuel stream.

Fig 9 illustrates combustor operation in the premix mode, meaning that fuel and air are mixed upstream of the burner. Premixing is conducive to lean fuel/air mixtures which minimize flame temperature and the formation of so-called thermal NOx. Of the three mechanisms associated with NOx formation—the other two being prompt NOx and the reaction that converts nitrous oxide (N2O) to nitric oxide (NO)—thermal NOx is the dominant mechanism in most GT combustors.

Note that sub-pilot fuel—approximately 1% of total fuel flow—is bled off the secondary fuel line and its flow rate is not controllable. The purpose of the sub pilot is to create a small-scale diffusion flame for stabilizing the premixed secondary flame which, in turn, allows the premixed primary flames to ignite.

Cohen stressed that the DLN-1 combustor designed for 9-ppm NOx demands very tight control of the fuel/air ratio over the entire load range, much tighter control than is required for the older 25-ppm-NOx DLN-1. Two terms to remember, he said, are these:

• Lean Flammability Limit, or the point at which the fuel/air ratio is too lean (that is, too low in value) to support combustion.
• Equivalence Ratio, which is the actual fuel/air ratio divided by the stoichiometric fuel/air ratio. DLN combustors operate very close to the Lean Flammability Limit and have a very narrow Equivalence Ratio operating range.

Cohen continued the “primer” portion of his presentation by re-familiarizing attendees with 11 air flow paths associated with the DLN-1. One look at Fig 10 and all the avenues for air entry into the combustor, each with its own purpose, illustrates why it is so difficult to tune these systems and keep them operating on-target.

Consider liner dilution holes, for example. The total area of all dilution holes—less for the 9-NOx than for the 25-NOx DLN-1— “tunes” the amount of air used for head-end premixing. Good penetration of dilution air is needed to achieve good mixing and CO burn-out. Fewer holes of large diameter are more effective for controlling CO than more holes of smaller diameter. Cohen mentioned that the 9 NOx system has three dilution holes, the 25 has four. Dilution holes may be of unequal size to achieve a proper exit-temperature profile when staged dilution is used.

Staged dilution is designed into transition pieces (TPs) for the 9 NOx. It consists of two 1-in.-diam holes on the inner panel of the TP about 9.5 in. from the exit. Staging here separates dilution into two zones for the purpose of improving CO burnout without affecting NOx emissions. Specifically, what it does is increase gas temperature in the TP by about 40 deg F for 10 milliseconds.

Primary/secondary fuel split. Next, Cohen displayed two figures—one for the 9-NOx combustor (Fig 11), one for the 25-NOx—to illustrate the emissions trends versus fuel-split characteristics common to both. Specifically, a minimum-NOx split always exists, and CO continuously decreases with increasing split. Furthermore, the split that achieves optimal NOx for a given combustor is not necessarily the split that produces the optimal CO.

Basics in hand, Cohen was ready to show users how to troubleshoot variations from the norm in fuel/air ratio that often are experienced after a combustor inspection (CI) or other outage. When such variations throw emissions out of compliance you have to move quickly, and in an organized manner, to identify and correct the root cause of the problem to assure continuity of operations.

First step is to plan your troubleshooting initiative. Look for the cause of variations in fuel/air ratio in the following places, he suggested:

• Overall flame zone. Check turbine firing temperature, combustor air flow distribution, and/or turbine air flow distribution.
• Can to can. Check for variations in fuel/air ratio among the combustor cans. Anomalies often can be traced to variations in the effective fuel-nozzle area—which is an important reason to flow-test fuel nozzles regularly.
• Within each combustor. A variation in fuel split often is the root cause here.

Cohen jumped ahead a bit telling the users that when called to plant experiencing high emissions following a CI, first thing he checks is the fuel split. Next, he looks at the exhaust spread, where a variation from the norm probably would indicate there’s a fuel-nozzle problem.

Liner inspection, testing. Then he outlined a game plan plant personnel could use to guide their troubleshooting efforts. Liner inspection was at the top of the list. Cohen said the liner refurbishment process can alter the effective area of dilution and cooling holes. For example, poor quality control in applying the thermal barrier coating (TBC) to the liner head can reduce the area of film cooling holes, which are only about a tenth of an inch in diameter. Look also for dings or TBC on the edges of the dilution holes; edges must be sharp.

Liner flow testing normally is not done, he continued, but it is something to consider when writing a specification for refurbishment. However, simply running a test is not a recipe for success. To illustrate the point, Cohen put up a slide that reviewed a flow test by a repair vendor that did not properly evaluate test results. In this case, CO emissions were above the permitted limit because of high excess air in the head-end premixer and too little excess air in the flame zone.

The vendor’s error was to assume that because the percent variance in total hole area for each can, compared to the average for all 10 cans, was less than the specified “allowable percentage” everything was fine.Not true in this instance because the measured area of the holes after refurbishment was 13% greater than the as-received measurement, allowing more air than planned into the head end and increasing CO.

Barometric pressure has a very significant impact on how your turbine fires and on overall fuel/air ratio.Proper laboratory calibration of barometric pressure transducers is important. Do not try to save money by calibrating yourself using airport or Internet information. These values generally are altitude-corrected and will read about 30 in. Hg. For example, the Denver airport most likely would tell you that the barometric pressure there is 14.7 psia (or 29.92 in. Hg)—the sea-level value— when, in fact, it’s really about 12.2.

Rule of thumb: For each 500 ft above sea level, a control system using sea-level barometric pressure will increase turbine inlet temperature by 10 deg F. Overfiring increases NOx production and also shortens the lifetimes of HGP components.

Delta p. Regarding errors in measuring inlet pressure drop, Cohen had this to say: Many operators are not aware that inlet delta p is an input to that part of the control system responsible for regulating turbine firing temperature; also, that pressure drop is measured by a single transducer—no triple redundancy here. Note that a delta p of zero is transmitted to the control system if the transducer fails. That typically results in a 5 deg F increase in firing temperature.

CDP. Errors in measuring compressor discharge pressure (CDP) are less common than they are measuring barometric pressure and inlet delta p. Most operators understand the importance of CDP and calibrate the triple-redundant transducers regularly.

However, Cohen cautioned that a system calibration is far more meaningful than a standard instrument calibration using a digital voltmeter (DVM). Reason is that control system cards drift over time. You need two people for a system calibration: one in the field to apply the test pressure and the other in the control room to read the output on the HMI (human/ machine interface).

Performance. Cohen recommended gathering performance data before and after every HGP and major target following an HGP or major, he continued, there are many more things to check than for the CI case. Generally whatever impacts turbine efficiency will impact fuel/air ratio. Examples include condition of HGP components, internal leakage rates, bucket tip clearances, etc. Without detailed test data it’s virtually impossible to pinpoint the reason(s) for deviation in fuel/air ratio in timely fashion.

Can-to-can. Regarding variations in fuel/air ratio from can-to-can, Cohen said one possible cause is plugged or partially plugged gas metering holes. The example he gave during this part of the presentation showed that while a large frame may look as indestructible as a heavy tank, its operation within permit limits dictates the treatment one might give a fine watch.

The example: Each dual-fuel primary nozzle for a DLN-1 has six metering orifices for gas or a total of 360 orifices per machine. At one site, a total of three orifices in two nozzles for one combustor can were partially plugged with debris (about the size of a small pea) from a primary purge during operation on liquid fuel.

Partial plugging of three of the 360 metering orifices caused CO to increase from 15 to 35 ppm. Flow data revealed that the effective orifice area in the affected combustor was only 3.2% lower than that of the 10-can average. A reasonable assumption: If one of 10 combustors was affected and engine CO emissions increased by 20 ppm, the “bad” combustor was producing about 200 ppm CO. Despite the dramatic change in operation of this combustor, there was no discernable cold spot in the exhaust temperature spread.

As Yogi Ber ra might have said: If you don’t know where to look or what to look for, you’ll never find it. No substitute for experience in plant operations and troubleshooting.

If you have 9 NO_x DLN-1 combustors, Cohen recommended a maximum of 1% (plus or minus) variation in the total effective area for both new and refurbished fuel-nozzle assemblies. More than that on primary nozzles, he said, and you might not be able to meet emissions expectations. This spec should be relatively easy to achieve for primary nozzles, more difficult for secondary nozzles.

Flow-test fuel nozzles before disassembly, Cohen advised. Identify nozzles with area variations significantly larger than the norm. If any, disassemble and inspect with the goal of pinpointing the cause and correcting it. Post-assembly flow testing is a given. Make sure all nozzles meet the area target.

Keep in mind that the presence of liquids in the gas stream is a common cause of nozzle fouling. Most gas cleanup systems are designed based on the gas supplier’s fuel spec, which often is inadequate for GT operations because of unexpected liquid slugging. If your nozzles suffer abnormal wear and tear and/or if hydrocarbon slugging is apparent, install an onsite gas pretreatment skid designed to your specific requirements.

In addition to liquids and solids (such as pipe scale and rust) that may accompany the gas you have under contract, review maintenance practices to be sure debris is not falling into open fuel or purge lines during outages; also, that there is no carryover of lube oil from an onsite gas compressor.

Options for preventing, eliminating varnish in hydraulic, lube-oil systems

Lubricant varnish continues to be a topic of great interest at gasturbine (GT) user-group meetings. Reason: It is the primary cause of the servo-valve sticking/seizing in control circuits blamed for many starting problems and turbine trips.

One of the first presentations to this industry segment on the subject was by ISOPur Fluid Technologies Inc’s (Pawcatuck, Conn) Chuck Mitchell at the 2004 meeting of the 501D5/D5A Users in Hartford. Mitchell’s objective was to raise awareness regarding varnish and why it occurs. Obviously, he had a solution to the problem.

Mitchell stressed the importance of eliminating fine particulates from hydraulic and lubricating oils in systems equipped with standard filtration equipment. Conventional filters, he said, were effective for removing particles 10 microns and larger; fine filters could extend that coverage down to about 3 microns.

However, Mitchell continued, particle- size analysis of representative lube oils suggested that roughly half of the particulates present ranged in size from 0.1 to 5 microns. Given that clearances can be 1 micron in loaded bearings, many of the particles escaping removal by standard filters could wedge between the shaft and journal and do damage.

The ISOPur solution, he explained, relies on Balanced Charge Agglomeration ™ (BCA), which “grows” small and sub-micron particles to filterable size so they can be removed by existing filters in the system—thereby reducing wear and eliminating the source of varnish. Mitchell seemed to initiate a flood of presentations on varnish and how to deal with it.

A frequent participant in user-group meetings has been Greg Livingstone, formerly of Analysts Inc, Torrance, Calif, and now with Calgary-based EPT Inc. Analysts developed the QSA™ (quantitative spectrophotometric analysis) test to determine the presence or likelihood of sludge and varnish buildup on critical components; EPT offers filters and other solutions to remove contaminants from lube and hydraulic oil as well as related services.

Livingstone says a primary cause of lubricant varnish is auto-degradation, which he defines as the creation of soft contaminants in a static body of oil—such as a shut-down lube-oil system serving a cycling or peaking GT. Soft contaminants, he continues, often are more troublesome to remove than the hard particulates on which lubricant experts traditionally have focused.

Livingstone adds that varnishpotential tests—such as QSA—alone will not tell you if auto-degradation is occurring, though such tests remain essential to your overall oil-condition monitoring program. Similarly, electrostatic separators, seen by many as a one-step “cure-all,” will not eliminate the problem. Note that the term “electrostatic separators” as used by Livingstone includes electrostatic oil cleaning, BCA, electrostatic filtration, etc.

To fully understand if your lubricant is undergoing auto-degradation, he continues, you need to assess the antioxidant health of the fluid and examine it for specific types of degradation byproducts. If auto-degradation is identified, you’ll need a holistic approach that includes monitoring of initial oil quality, analysis and additive replenishment for inservice oil, and the installation of appropriate oil-cleaning technology to remove existing varnish and slow the degradation process.

More background on lube-oil testing and quality improvement is available in the following articles accessible through www.combinedcyclejournal. com/archives.html: Summer 2004, click on cover “Maintain lube oil within spec to ensure high reliability”; 3Q/2005, click “The lowdown on the sticky subject of lubricant varnish; 3Q/2006, click “Gasturbine valve sticking. . .the plot thickens.”

The 7EA Users Group has identified varnish and other lube-oil issues as an area of significant interest to its membership. It should be. There are more than 1000 Frame 7s (model As through EAs) in operation worldwide, 70% of those in the US. About 60% of the total population is used in peak-power applications, meaning the majority of the units in the fleet are particularly susceptible to varnish formation based on Livingstone’s experience described above.

There was a varnish-related formal presentation at the organization’s 2006 meeting in San Diego.A representative of Pall Corp, Port Washington, NY, brought the group up to date on a new filter media designed to minimize the potential for electrostatic discharge in hydraulic, lubricating, and fuel systems.

Recall from the references suggested above that electrostatic spark discharge (ESD) from filters has been observed and documented in several powerplants worldwide. It is described this way: As oil flows through the small openings of a filter, molecular friction is produced and it creates static electricity. When the electrical charge in the fluid accumulates to a given point, the energy is released in the form of a spark, arcing from the sharp edges inside the filter housing (Fig 12).

The locally high temperatures produced by ESD oxidize the oil; the byproducts of this oxidation include varnish. It follows then that a filter medium capable of limiting ESD would have a positive impact on oilsystem health.

The speaker explained that the potential for electrostatic charging increases with decreased conductivity, increased flow rate or velocity, and the additive package. Also that fluid conductivity—which helps with charge dissipation—increases with temperature (lower viscosity), water content, additive concentration, and the amount of dust and other impurities in the oil.

Next, he described Pall’s test setup for measuring electrostatic charge, explained the charge collector, and presented the characteristics of the four oils tested. Regarding the last, the products evaluated were one turbine lube oil, two commercial hydraulic oils, and a hydraulic oil for the military. Additive packages included R&O, antiwear, and antiwear/antioxidant. Viscosities varied from 14 to 47 centistokes, dielectric strengths from 15.5 to 27.3 kV, and conductivities from 39 to 1460 picoSiemens/ meter (pS/m).

Results were presented as average charge generation in nanoamps for three filter materials: standard glass fiber, surface-modified glass fiber, and Pall’s new glass-fiber-based ESD. One set of tests was run on these materials after heat-soaking at 300F for one hour; a comparable set of tests without heat soaking prior to use. All tests were conducted with the oils at ambient temperature.

Here are the results of greatest interest to plant personnel:

• Charge generation for the standard and surface-modified glass fiber materials was about two times greater after heat soaking. By contrast, charge generation for Pall’s new filter media was the same whether heat-soaked or not.
• For the heat-soaked samples, charge generation for the new ESD media was a factor of 15 less than that produced by the standard glass fiber and six times less than that produced by the surface modified glass fiber media.

Field trials supported the test results. In sum, the new Pall filter substantially reduced charging—and eliminated all signs of noise, sparking, and filter damage—both in the laboratory and in field tests. Specifically:

• In a manufacturing plant ’ s hydraulic system, the new filter media lowered the charge generated to a negligible amount and eliminated both noise and sparking.
• In an injection molding hydraulic system, the new filter media eliminated noise and burn marks and reduced the charge generated by about 75%. It did the same in a paper-mill hydraulic system except that the reduction in charge produced was 98%.
• In a powerplant lube-oil system a distinct clicking noise that was apparent before the change in filter medium disappeared.

In his conclusions, the speaker said that electrostatic charging can be a problem in hydraulic and lube-oil systems using any manufacturers’ standard glass-media filter—although it occurs relatively infrequently. Also, that grounding housings and pipes do not reduce the charge generated.

The editors followed up with the filter experts at Pall following the San Francisco meeting and learned that the company’s electrostaticdischarge- resistant filter media is now available commercially in various cartridge configurations and in several porosity grades. A company spokesperson said, “These filters have been employed in various industries and applications and have a track record of resolving the tough problem of electrostatic discharge and its associated damage while providing highly reliable fine filtration.”

Alternatives for varnish mitigation

A feature of the 7EA Users’ 2007 conference was a three-vendor panel describing alternative solutions for preventing varnish formation and for clean-up of existing deposits. It was developed by Julie Turner, plant manager of Progress Energy Florida Inc’s Intercession City facility.

The editors believe this was the first time a user group provided owner/operators the opportunity to compare the various offerings on a level playing field. Presenters were ISOPur; the Hilco Div of Hilliard Corp, Elmira, NY; and C C Jensen Inc, Tyrone, Ga. These companies, plus the participation by Pall last year and the availability of Kleentek Inc (Cincinnati) and EPT personnel at the vendor fair in San Francisco, allowed 7EA users to access information first-hand on perhaps all of the leading commercial varnish solutions.Analysts Inc and Chevron Lubricants were at the vendor fair as well to answer questions on test procedures and lubricant properties.

Mike Long, Hilliard’s product engineering manager, focused his presentation on the elimination of varnish root causes rather than its removal after formation. You knew where Long was headed from the get-go when he said, “Static discharge is not a fluid problem and not a cartridge problem. Its root cause is the use of API (American Petroleum Institute) Type II lubricant base stocks and low fluid conductivity—less than 35 pS/m.”

Long added that traditional staticdischarge control techniques—such as use of conductive filter elements or of large-diameter filter elements to reduce velocity through the screen and lower fluid shear—are not the complete answer because they do not address low fluid conductivity. Then he introduced his company’s new anti-static element for lube/ hydraulic-oil conditioning, which Long said was capable of raising fluid conductivity above 200 pS/m (Fig 13). It is designed for kidney-loop service.Note that fluid conductivity determines when to replace filter elements of this type, not pressure drop.

Next he discussed microdieseling, which contributes to varnish formation. It is caused by dissolved gases in the lube oil—mostly nitrogen and oxygen. When gas bubbles transition from a region of low pressure to one of high pressure, the gases implode, generating sufficient heat to thermally degrade the fluid. Oil analyses from three F-class machines from different areas of the country showed similar dissolvedgas compositions. A vacuum dehydrator/degasser removes both moisture and dissolved gases (Fig 14).

Thus an effective system for preventing varnish formation would combine an anti-static filter element and dehydrator/degasser. A threeweek trial of an F-class kidney loop equipped with both the dehydrator/ degasser and anti-static filter element produced these dramatic results:

• Fluid conductivity increased from 19 to more than 500 pS/m before it began drifting backward. The parameter is measured in-situ by a digital conductivity meter that meets the ASTM D2624 test standard.
• Moisture content of the oil was reduced by 80%.
• Dissolved gases were reduced by more than 50% as confirmed by a third-party laboratory.
• Improvement in ISO-4406 cleanliness codes from 20/18/15 to 18/16/14

Long estimated the cost of the varnish prevention system described at somewhere between $25,000 and $50,000 depending on throughput. Annual operation and maintenance— including electricity and consumables (filter elements, gaskets, etc)—would be less than $5000, assuming quarterly filter replacements.

ISOPur’s David Cummings told the group that key to preventing varnish issues are a good oil supplier, good filtration system, proactive user, and a good laboratory. Regarding filtration, he said, the BCA improves filtration efficiency by making both hard and soft particles larger (Fig 15), plus it prevents varnish buildup and removes existing varnish. System effectiveness is illustrated by Fig 16, which shows how average particle size increases with each pass of the oil through the kidney loop.

A free-standing oil conditioning skid that would be piped into a kidney loop off the main lube-oil reservoir is shown in Fig 17. It consists of a prefilter, charging/mixing unit, collection filter, and variable-speed gear pump. The ultra-clean oil produced acts as a solvent and pulls back into the oil the sludge and varnish hiding out in servos, gearboxes, sumps, etc.

Cummings (dcummings@isopur. com, 860-599-1872) agreed with Livingstone’s comment above that peaking systems do create a more difficult environment for varnish removal/ control. All of the conditions that create the precursors to varnish can increase when the turbine is on turning gear, he added. Time on turning gear and the level of antioxidants in the oil impact varnish production, removal, and control, he continued. For best results, the BCA system should remain in operation when the unit is on turning gear to remove oxidized material, extend oil life, and minimize varnish.

ISOPur conducted 19-week BCA performance tests in parallel on seven GE 7FA engines equipped with the system at Tampa Electric Co’s Bayside Power Station. Each of the units has a 6000-gal main lube-oil tank, meaning the 10-gpm kidney loop provides about 2.5 reservoir “turns” daily. Average varnish-potential rating dropped from 38 to 23 during the period, with the range of unit “end of test” VP ratings extending from 8 to 33. VP numbers at the start of the test were between 32 and 42.

Gravimetric analysis numbers were more tightly bunched at the end of the test—between 0.18 and 0.32 and averaging 0.25—after starting in the 0.4 to 0.6 range with 0.5 as the average. Total count of 0.2- to 2.0-micron particles averaged 200,000 at the start of the test program and all but one of the units (test stopped early) finished the program at 25,000. Likewise, water content of the oil averaged 40 ppm at the start and all but one unit with suspect numbers ended at about 10 ppm.

Cummings recommended that filter elements be inspected and replaced at frequent intervals. He offered a change-out plan for both prefilters and collection filters for older (used) oils and new. For units with used oil, your total Year One expense will be equipment capital cost and about $1650 in replacement filters; for new oil, the capital cost is the same but replacement filters should not cost more than about $1000 the first year. Replacement filters for both new and old oil every year after the first will run about $700.

Testing should include submicron particle count/distribution; VPI or QSA; FTIR (Fourier Transform Infrared Spectroscopy) to evaluate an oil’s condition and the presence of contaminants—such as water; and the so-called RULER test, to measure the concentration of antioxidants present in the oil—primarily phenols and amines. Based on initial findings, an ongoing retest program can be developed.

Cummings closed by saying that BCA technology has been validated by GE Energy and other OEMs. Specifically, GE TIL 1528-3 (Nov 18, 2005) stated, “GE has performed extensive studies to validate the use of BCA technology. . .this technology can be used to mitigate a current varnishing issue or to prevent the occurrence of it.” System components have GE part numbers and can be ordered online through GE PartsEdge.

Justin Stover, C C Jensen’s sales manager, closed out the program with a presentation on the value of cellulose filter media for adsorbing varnish. C C Jensen, a 50-yr-old company with Danish roots, is a relative newcomer to the US electric power industry. However, its filtration systems are particularly well known in the global marine and oil and gas industries. More recently, several manufacturers of wind generation systems have standardized on C C Jensen filtration packages for their gear-oil and hydraulic pitch-control systems. Tens of thousands of these currently are in service worldwide.

Stover began with the basics, including a review of adsorption physics. Recall from your formal education that adsorption is all about using solids to remove specific substances from gases and liquids; molecular attraction is what makes the process of absorption work. Specific to this discussion, when varnish passes by an adsorbent, it attaches to its surface (Fig 18).

Cellulose is particularly effective in this regard; its high polarity is well suited to attracting oxygenated molecules—such as varnish. Stover stressed that this was a “natural” process—no voltage required, no control system, etc. Capacity is determined solely by surface area. He said that just one gram of cellulose has a surface area of about 4000 ft2 and that a standard filter cartridge contains 3600 grams of cellulose; you do the math.

Exactly what happens inside the filter media is described in Fig 19. Here’s a more detailed explanation of the terms used in the drawing: Diffusion is the transport of matter (varnish in this case) from one point (the oil) to another (the filter media). Film diffusion describes how the varnish molecules are drawn to the boundary of the cellulose fiber by means of the inherent physical forces (polarization, electrostatic, and hydrogen bonding).

Once “inside,” the varnish molecules move among, or between, the cellulose molecules in open spaces. The spaces are large relative to the size of the molecules, hence the term macropore diffusion. Next, the varnish molecules come to rest on the adsorbent surface—that is, they diffuse from the fluid onto the cellulose molecule (micropore diffusion).

Stover said the filtration system is easy to operate and maintain and that it is installed in a kidney loop like the other offerings described above (Fig 20). To illustrate performance on a 7EA, he used centrifuge samples and color values of the oil taken between September 2006 and March 2007 (Fig 21). The color value at the beginning of the test was a 63. More specifically, the number of particles in the size range of 0.2 to 1 micron was more than 20 million. At the end of the test, color was 0 and the particle count was less than 3400.

The capital cost of a fixed filtration system serving a 7EA in a typical low-varnish environment is about $7000. Annual filter costs are a nominal $1000 for peaking turbines; less than half that for unit in base-load service. ccj