By: Brad Buecker, Contributing Editor
The threat of flow assisted corrosion in power plant equipment demands close attention to proper water chemistry control.
Figure 1. Typical FAC Pattern. Photo courtesy of GE Betz.
Flow assisted corrosion (FAC), sometimes called flow accelerated corrosion, is a serious issue at power generating facilities, as dramatized in recent years by a number of fatalities due to FAC-induced failures. The FAC threat to personnel safety and to plant performance demands close attention to proper water chemistry control. This article outlines current practices regarding boiler feedwater treatment, especially as they relate to FAC prevention.
FAC in Conventional Boilers
Classical steam generation chemistry teaches that when a boiler is first fired, and where the feedwater and boiler water contain little or no dissolved oxygen, carbon steel feedwater piping and boiler tubes develop a protective layer of magnetite (Fe3O4), while copper alloy heat exchanger tubes develop a protective layer containing cuprous oxide (Cu2O). Throughout much of the last century, utility chemists were taught to completely remove oxygen from feedwater to prevent oxidation of these protective layers and minimize corrosion. So, typical fossil plant design often included an air removal system in the condenser, a deaerator (DA) for additional air and non-condensable gas removal ahead of the high-pressure feedwater heaters, and oxygen scavenger injection to feedwater to remove any traces of oxygen that might escape the DA. Typical treatment programs consisted of injecting ammonia or amine into the condensate to raise the pH to 9.2-9.6 for all ferrous systems and 8.8-9.1 for mixed metallurgy systems, and injecting hydrazine or one of the specialty organic scavengers into the feedwater. Plants commonly treated feedwater to less than 1 ppb of oxygen with 20-100 ppb of the oxygen scavenger.
While it is well documented that dissolved oxygen can induce serious boiler and feedwater system corrosion, especially during unit shutdown, the complete removal of oxygen during normal operation can be very problematic, and has led to several sudden failures of mild steel by flow assisted corrosion. FAC develops at flow disturbances, e.g., feedwater and economizer elbows, reducers, and tees, in strongly reducing environments. To understand the problem, first consider the nature of the protective magnetite layer. The compound is actually a joint mixture of FeO and Fe2O3 that often exhibits a rippled or irregular pattern. Iron exists in a +2 (ferrous) oxidation state in FeO and +3 (ferric) in Fe2O3. The ferrous ions are those that are susceptible to FAC, and in affected zones these ions migrate out of the magnetite matrix.1 The reducing environment continually regenerates ferrous iron, whose constant migration weakens the wall structure and eventually reduces pipe strength to the point of sudden failure. Figures 1 and 2 respectively illustrate a typical FAC pattern and the wall thinning that the mechanism can induce. At several plants in recent years, personnel have been working near equipment that catastrophically failed due to FAC. The sudden release of pressurized, extremely hot water that flashed to steam left no time to escape.
Boiler design modifications and improved materials have helped reduce the threat of FAC. Feedwater treatment has also evolved to meet the problem. Many once-through units in the world now use oxygenated treatment (OT) programs, in which oxygen is deliberately injected into the condensate and feedwater to produce a smooth, dense protective layer of ferric oxide hydrate (FeOOH). A dissolved oxygen range of 30-50 ppb is common in the U.S. Oxygenated treatment, when applied properly, can significantly lower steel corrosion rates and subsequent iron transport to the boiler. Popular practice with this treatment, in the U.S. at least, is to add a small amount of ammonia to the condensate to maintain a slightly alkaline pH between 8.0 and 8.5 to provide additional corrosion control. OT with an ammonia supplement is known as combined water treatment (CWT). An absolute requirement for OT is high purity feedwater, and condensate polishers are a must, as otherwise contaminants and deposits will generate severe localized oxygen corrosion. Also, OT is not applicable in units containing copper-alloy heat exchangers, as oxygen and ammonia in combination would rapidly corrode the tube metal.
Oxygen for OT treatment programs can be supplied in several different ways. One plant’s OT system utilizes simple oxygen cylinders manifolded together to provide a long lasting supply. At another plant, liquid oxygen (LOX) is the feed supply. Plant personnel order bulk containers of LOX on an as-needed basis. Several control mechanisms are possible for oxygen feed to OT units. Flow may be adjusted manually or controlled with a feed-forward flow signal and dissolved oxygen concentration feedback loop.
But what about drum boilers? Oxygenated treatment has been successfully employed in perhaps 50 or so drum boilers throughout the world, but many still continue to operate on the ammonia/oxygen scavenger technique for feedwater treatment. This is especially true in systems with copper-alloy condensers and/or feedwater heaters. The presence of copper puts plant chemists between a rock and a hard place in trying to prevent FAC while minimizing copper corrosion. This is particularly true in units that operate at and above 2400 psi, where copper carryover to the turbine becomes an issue. A few ppb of oxygen will maintain a protective oxide layer on carbon steel without causing serious copper degradation, but control has proven very difficult, requiring precise monitoring and feed of the oxygen scavenger to remove most but not all oxygen.
Attempts to control oxygen scavenger feed based upon signals from scavenger and/or dissolved oxygen monitors have not always been successful. An emerging technique for fine control of oxygen scavenger feed is oxidation-reduction-potential (ORP) monitoring. Feedwater treated with an excess of oxygen scavenger exhibits an electrochemical potential that may be as low as -350 mV (as measured with a platinum electrode vs. a silver-silver chloride reference electrode). Conversely, the electrochemical potential of feedwater on OT will range from +50 to +150 mV. The key with mixed metallurgy systems is to find an ORP range in between that minimizes both copper and iron corrosion. The range varies depending upon the type of copper alloy within the system and the location of copper alloy heat exchangers. The following discussion highlights practical results obtained by Steve Shulder, Chemistry Consultant, and colleagues at Constellation Power.
The personnel at Constellation Power have had to deal with many different issues. One feedwater system, on a 2750 psig unit with full-flow condensate polisher, has a copper-nickel alloy condenser (90/10 Cu-Ni in the main section and 70/30 Cu-Ni in the air removal compartment) with carbon and stainless steel tubed feedwater heaters. This unit had previously suffered a loss of turbine performance due to copper corrosion and transport that had been caused by elevated levels of dissolved oxygen. An automatic system now controls the hydrazine pump injection rate to maintain the ORP around -100 mV with an appropriate hydrazine residual in the range of 5-30 ppb.
In a 2050 psig unit with a titanium condenser, a carbon-steel low-pressure heater followed by an Admiralty-tubed (71% copper, 28% zinc, 1% tin) low-pressure heater, and no high-pressure heaters, ORP is used as a monitoring device for manual hydrazine feed adjustment. The optimum ORP range is 0 to -100 mV. The unit cycles many times during a year, which makes chemistry control difficult. The plant staff found that ORP monitoring has allowed them to fine-tune chemical feed and reduce corrosion product transport to the boiler. And, in a third load-following unit that operates at 2000 psig with an arsenical aluminum brass condenser (90.5% copper, 7% aluminum, 2.5% iron), two 70/30 copper-nickel low-pressure heaters and two, stainless steel high-pressure heaters, the staff manually adjusts hydrazine feed based on an optimum ORP range of -250 to -350 mV.2
Personnel at Western Farmers Electric Cooperative’s Hugo Plant in Ft. Towson, Okla., have adopted on-line ORP as a measuring and control technique for feed of the specialty organic oxygen scavenger methyl ethyl ketoxime.3 Because ORP measures the potential of solution regardless of the oxygen scavenger, the technique offers a more flexible analytical choice. An important step to establish the proper ORP range involves correlating the readings with other measurements. Any change in feedwater treatment should be conducted in tandem with frequent and regular iron and copper analyses and corrosion product transport monitoring to allow the chemist to fine-tune the ORP control range and measure the impact of the treatment change.
Figure 2. Typical FAC-induced Wall Thinning. Photo courtesy of GE Betz.
These examples clearly illustrate the variability in ORP control ranges and applications. ORP monitoring has not been universally perfected yet, and some plant chemists do not use ORP for oxygen scavenger feed control due to instrument reliability and accuracy issues. However, ORP’s potential (pardon the pun) for becoming a primary oxygen scavenger control method is pushing instrument manufacturers toward quick solutions of many problems.
FAC in HRSGs
Flow assisted corrosion is of concern in HRSGs because of boiler design. A common HRSG configuration is the multi-pressure, vertical-tube arrangement. The tubes in these steam generators are relatively short in height with sharp radius elbows at header and drum connections. Thus, they are prime locations for both water- and steam-induced FAC.
Also, in many HRSGs the low-pressure (LP) evaporator serves as the feedwater heating source for the intermediate and/or high-pressure evaporator. The LP evaporator may be equipped with an integral deaerator. This arrangement precludes the use of solids-based boiler water treatments such as phosphate or caustic. A common chemical program for the LP circuits is all-volatile treatment, which of course raises the specter of FAC. The ORP monitoring and control techniques mentioned earlier are a possibility for this application. Constellation has incorporated hydrazine pump control using the plant DCS with feedwater flow and ORP measurements to minimize high concentrations of hydrazine in the LP evaporator of a recently commissioned combined-cycle plant. Another possibility is oxygenated treatment of the feedwater, as combined-cycle HRSGs typically do not have feedwater heaters and thus may not have any copper in the system.
FAC prevention is possible during HRSG manufacture. A small amount of chromium alloy in steel greatly inhibits FAC. Molybdenum is another beneficial alloy. An increasingly popular design technique for HRSGs is to fabricate waterwall elbows and other FAC-susceptible components with steel containing a percent or so of chromium. This proactive approach is an option that future plant owners should strongly consider.
Inspecting for FAC
Many plant personnel have operated units for years on conventional oxygen scavenger chemistry. All the while, FAC may have been eating away at feedwater and economizer elbows and other flow disturbances. Plant management at any of these facilities should consider non-destructive wall thickness testing to determine if FAC is indeed a problem. The fact that FAC has caused several fatalities should be more than enough motivation to inspect for damage.
FAC is an extremely serious issue, whose importance cannot be overemphasized. In fact, the Electric Power Research Institute (www.epri.com) has published a book on the subject which anyone, regardless of whether their company belongs to EPRI or not, can purchase. The book also discusses two-phase FAC, which can occur in pipes carrying water/steam mixtures.
- Daniels, D., “HRSG Failure Mechanisms – Waterside,” Proceedings of the 22nd Annual Electric Utility Chemistry Workshop, Champaign, Illinois, May 7-9, 2002.
- Shulder, S., “Practical Applications of Oxidation Reduction Potential (ORP) to Control Oxygen Scavenger Injection in Fossil Power,” Proceedings of the 21st Annual Electric Utility Chemistry Workshop, Champaign, Illinois, May 8-10, 2001.
- Pike, T., “An Improved Method for Monitoring Low Concentrations of Volatile Oxygen Scavengers,” Proceedings of the 62nd Annual Meeting of the International Water Conference, Pittsburgh, Pennsylvania, October 21-25, 2001.
Brad Buecker is the plant chemist at Kansas City Power & Light Company’s La Cygne, Kansas power station. He has previous experience as a chemical cleaning services engineer, a water and wastewater system supervisor, and a consulting chemist for an engineering firm. He also served as a results engineer, flue gas desulfurization (FGD) engineer, and analytical chemist for City Water, Light & Power, Springfield, Ill. Buecker has written more than 50 articles on steam generation, water treatment, and FGD chemistry, and he is the author of two books on steam generation chemistry published by PennWell. Buecker has a BS in chemistry from Iowa State University, and is a member of ACS, AIChE, ASME, and NACE. He can be reached at [email protected]