By Brad Buecker, Contributing Editor
The steam surface condenser is an integral component at electric utilities and other facilities that generate power. It is also the point where highly pure condensate can potentially come in contact with very impure cooling water via a condenser tube leak. The consequences literally can be catastrophic for plant operations.
During much of my career at two coal-fired utilities, I monitored condenser performance for a number of units, where the condenser tube metallurgy ranged from Admiralty brass to copper-nickel alloys to stainless steel. This article outlines many of the factors that can influence condenser tube corrosion and the effects contaminant in-leakage can have on steam generating units.
Figure 1 Hydrogen damage in a waterwall tube
Photo courtesy of the Nalco Co. Click here to enlarge image
The literature contains many references to the problems caused by cooling water infiltration into steam system condensate. We will examine a few of these problems as a refresher on how impurities can affect boiler water chemistry.
Numerous authors have described water as the closest thing to a universal solvent. While this is an oversimplification, water’s unique properties do give it powerful solvating characteristics. Typical cooling waters, whether from a surface supply or from wells, contain many compounds absorbed from the earth and the atmosphere. Ground waters usually contain greater mineral concentrations than surface waters, including calcium, magnesium and iron. These dissolve in water as it percolates through the soil and underlying rock formations. Surface waters are often softer, but contain many more suspended solids as well as organic compounds, the latter due to decomposing leaves and plant material. Common organics include tannins and lignins, whose rather complicated structures resemble the carbon-based plant material from which they derive.
Figure 2 MIC in a stainless steel tube
Source: The Nalco Guide to Cooling Water Systems Failure Analysis. Click here to enlarge image
When contaminants enter a steam-generating system, the high temperatures will initiate many reactions. Well known is the potential for scaling by calcium salts. Silica may precipitate by itself to form a hard, insulating deposit. Or it may combine with magnesium or other ions to form additional troublesome scales. Impurities, particularly silica, will carry over with steam and precipitate in turbines and reheater lines. Although boiler water treatment programs help mitigate these reactions, if an upset is too severe the treatment may be overwhelmed, as the following lesson illustrates.
Lesson No. 1
An 80 MW unit supplied by a 1,250 psig coal-fired, cyclone boiler had just been returned to service from a scheduled outage. Laboratory personnel discovered a condenser leak was allowing contaminants to enter the system, such that condensate total-dissolved-solids (TDS) concentrations at times reached 0.75 ppm. Although lab staff asked that the boiler be taken off line immediately, operations managers declined due to load demand issues. The boiler was on phosphate treatment, so lab staff increased monitoring frequency and attempted to maintain phosphate and pH levels within recommended guidelines. After roughly three weeks, operators discovered the source of the leak and corrected the problem.
Two months later, boiler waterwall tubes began to fail with alarming frequency. The unit came off numerous times for tube repairs. In at least one instance it had only been back on-line for a few hours when another tube failed. The failures happened so regularly that plant management scheduled an emergency tube replacement during an upcoming outage at a cost of over $2 million. The mechanism attributed to these failures was under-deposit corrosion caused by excessive sludge and scale formation. Interestingly, the leak was not from a failed condenser tube. The condenser hotwell is equipped with a drain line that discharges to the cooling water outlet tunnel. During the outage, an operator opened the line to drain the hotwell but forgot to close the isolation valve before startup. Once the unit went on-line, the strong condenser vacuum pulled cooling water into the hotwell.
Impurities can quickly initiate corrosion in a boiler. The reaction below outlines what occurred in this case history.MgCl 2 + 2H 2O + heat → Mg(OH) 2↓ + 2HCl
Magnesium chloride or sulfate salts react with water to produce a magnesium hydroxide precipitate, plus acid. This can cause the boiler water pH to fall well below neutral, whichgiven the harsh conditions inside a boileris potentially catastrophic. Not only will acid cause general corrosion, but the hydrogen atoms produced by the reaction form molecular hydrogen, which penetrates tube metal due to the element’s small size. The hydrogen gas by itself induces internal pressure in the steel that can cause metal cracking. It will also react with carbon to produce methane (CH4), a larger molecule, which exacerbates the cracking mechanism.
Hydrogen damage failures may result in a matter of days without any appreciable metal loss following a significant condenser tube leak, particularly if the cooling water is cycled up in a cooling tower or comes from brackish or ocean makeup sources. The Electric Power Research Institute (EPRI) recommends immediate boiler shutdown if the pH drops to 8.0. The following case history illustrates how rapidly contaminant in-leakage can affect boiler water chemistry.
Lesson No. 2
This problem involved a 200 MW unit with a 2,400 psig, coal-fired tangential boiler. The unit was on phosphate treatment with typical bulk water phosphate concentrations of 2 to 6 ppm. At the time of this event, the unit’s only on-line chemistry monitoring consisted of a sodium analyzer at the condensate pump discharge (CPD). The monitor was not equipped with an alarm. On the morning of the upset, a unit operator checked the monitor at 7:00 a.m. and found the sodium concentration measured less than one part-per-billion (ppb). When lab chemists checked at 7:45 a.m., the monitor needle was pegged beyond the upper limit of 100 ppb. Lab personnel took a boiler water sample and found that no phosphate remained in the sample and that the pH had dropped to 5.8. The chemists notified operations personnel, who took the unit off as quickly as possible. During this time the chemists injected trisodium phosphate into the boiler water to stabilize the pH and establish a phosphate residual. The operators opened the blowdown drains on the lower headers to remove solids.
Once the unit came off-line, maintenance personnel discovered the problem. A plug had fallen out of a topmost condenser tube, which had previously failed due to steam erosion. (Another failure mechanism to inspect for during outages.) The tube had massive failures across its length, so large quantities of cooling water poured into the condenser. Even though this condenser is on a once-through cooling system supplied with relatively soft lake water, the contaminants quickly consumed the boiler’s phosphate. Quick action by plant staff prevented serious scale formation or corrosion. The boiler was chemically cleaned at the earliest available opportunity and no tube failures occurred. However, the event helped convince utility managers to install a comprehensive on-line water chemistry monitoring system.
Lesson No. 3
A unit with a condenser tubed with 304 stainless steel came down for a scheduled maintenance outage, but the waterside of the condenser was not drained. Many of the tubes remained immersed in standing water for about a month. When the unit came back on line, cooling water in-leakage to the condensate was severe. Upon inspection, plant personnel discovered numerous pinhole failures in the tubes. The failure mechanism was microbiologically induced corrosion (MIC) produced by the stagnant conditions. When microbes attach to tubes, anaerobic bacteria will grow and flourish underneath the overlying layer of slime. Anaerobic bacteria do not, as the name implies, use oxygen in their metabolic processes. Rather, they use other chemicals such as sulfate (SO4) for energy. Resulting byproducts include acid and other compounds, including hydrogen sulfide (H2S). These compounds then produce pitting in the tube metal. Pitting is an insidious corrosion mechanism, as failures result with little loss of tube material.
Figure 3 Sulfide pitting of a copper-alloy tube
Photo courtesy of the Nalco Co. Click here to enlarge image
Underdeposit corrosion will also occur while units are in operation if scale or microbiological deposits are allowed to accumulate on condenser tubes. These deposits will also severely retard heat transfer, where loss of efficiency can cost a large plant thousands of dollars a day making it important to implement and monitor proper cooling water chemistry programs.
Lesson No. 4
Another condenser was originally equipped with Admiralty (70 percent copper and 30 percent zinc) tubes, which had begun to fail after 17 years due to steam-side ammonia/oxygen attack (discussed below). Plant personnel decided to replace the tubes with 90-10 copper-nickel, a material that had performed well in other condensers. Within two years, the new tubes began to fail. When a tube was pulled for examination, numerous pits were observable on its waterside. A single one-foot-long half-section sample contained eight through-wall penetrations. The corrosion was traced to the manufacturing process and a supplier that had used a lubricant containing sulfide (S-2), which is a corrosive agent to copper. The supplier did not clean the tubes before they were installed so the sulfide kept eating through the tubes until they failed.
Lesson No. 5
The condenser outlined in lesson 4 had operated well. But without warning numerous failures began to occur in the air-removal section. These failures caused a number of forced outages. During one outage, maintenance personnel pulled four tubes that previously had been plugged. Each of the four Admiralty brass tubes showed circumferential gouges at interface points between the tubes and tube support plates. The gouges showed many small cracks and one or more of the gouges in each tube contained a through-wall penetration. The corrosion was identified as ammonia-oxygen attack of the metal.
Air enters a steam-generating system at points around the condenser due to the strong vacuum generated within. Prime spots for in-leakage include the expansion joint between the turbine and the condenser, penetrations of heater drips lines into the condenser shell, turbine seals and explosion diaphragms and condensate pump seals. Air in-leakage is almost impossible to prevent, but the effects are manageable under normal conditions. Most condensers are equipped with one or more air-removal compartments in which a mechanical vacuum is applied to pull in gases from the condenser and exhaust them to the atmosphere. However, this process also concentrates chemicals in the air removal zone, including oxygen and ammonia, the latter generated by feedwater chemistry treatment. Concentrations of ammonia are typically high in condensate that collects on tube support sheets and flows downwards and around the tubes to the hotwell below. Oxygen converts the protective cuprous oxide (Cu2O) film on the tube surfaces to cupric oxide (CuO), after which the ammonia complexes the copper and puts it in solution.Cu +2 + 4NH 3 → Cu(NH 3) 4+2
Lesson No. 6
An interesting cooling water in-leakage problem occurred on one of the condensers outlined above. The condensate pump discharge (CPD) is equipped with on-line sodium and cation conductivity monitors. Sodium concentrations typically range between 0.5 and 0.9 ppb. During one three-week stretch of baseload operation, condensate sodium levels periodically increased to a range of 1 to 2 ppb, where they remained for anywhere from several hours to as long as a day before returning to normal. The sodium fluctuations could not be traced to operational factors such as load changes or soot blowing, so plant chemists concluded that a condenser tube leak had developed. At the earliest opportunity, operators took the unit off line. Maintenance personnel using the dye-check method found a small leak in one tube which, once plugged, made the problem disappear. However, the question remained as to why the contamination appeared and then disappeared with regularity. What seems certain is that the pinhole periodically plugged with debris that entered with the cooling water.
Lessons Nos. 7 and No. 8
In a 1992 Power Engineering article, I reported on a condenser performance program that proved excellent for monitoring heat transfer and troubleshooting problems.1 Subsequent work produced several lessons learned, two of which are outlined below.
Figure 4 Ammonia-oxygen attack of a copper alloy tube
Source: The Nalco Guide to Cooling Water Systems Failure Analysis. Click here to enlarge image
I had been performing thrice-weekly cleanliness factor analyses on a condenser rated at 1 million lb/hr of steam flow. The values remained very steady in the mid 70 percent range for several months, but dropped within two days to 45 percent. Waterside fouling does not occur this rapidly and such drastic changes are more indicative of excess air in-leakage. The maintenance department was notified and when they inspected the condenser they discovered a crack in the condenser shell where a heater drips line penetrates. Once maintenance workers sealed this crack, the cleanliness factors returned to previous values where they remained for another two months until suddenly dropping again. The seal had failed. This time the maintenance crew welded a collar around the drips line, which sealed the crack entirely and solved the problem.
In another instance, I had been collecting thrice-weekly readings on two, 690,000 lb/hr. condensers when one condenser suddenly began performing erratically. At maximum unit loads, the cleanliness factors ranged between 70 percent and 75 percent. At low loads, however, the factor dropped as low as 18 percent. Here, too, such fluctuations could not have been the result of waterside fouling. Utility managers brought in a leak detection firm to look for air leaks using helium leak detection to completely check the condenser and low-pressure end of the turbine. The inspectors classified leaks as large, medium and small, and found over a dozen leaks, including two large leaks (one of which was caused by a crack in the expansion joint between the turbine exhaust and condenser). Maintenance crews repaired all leaks, but this did not solve the problem.
Finally, an operator discovered that a trap on a line from the gland steam exhauster was sticking open at low loads. The trap and line are designed to return condensed gland steam from the condensate subcooler to the condenser, but vent gases to the atmosphere. When the trap stuck open, the strong condenser vacuum pulled outside air in through the vent. Once maintenance personnel replaced the trap, the condenser performance problems disappeared. This example illustrates how complex the task can sometimes be to identify sources of air in-leakage.
1. B. Buecker, “Computer Program Predicts Condenser Cleanliness”; Power Engineering, Vol. 96, No. 6, June 1992.
Brad Buecker is a contributing editor for Power Engineering.