How to identify, prevent waterside failure mechanisms in your HRSGs

By David G Daniels, Mechanical & Materials Engineering

None of the waterside fail­ure mechanisms found in heat-recovery steam generators (HRSGs)—including flow-accelerated cor­rosion (FAC), corrosion fatigue, thermal fatigue, and under-depos­it corrosion)—are exclusive to this type of boiler. However, failures attributed to these mechanisms have occurred more frequently at some combined-cycle plants than at fossil-fired steam/elec­tric stations. Even relatively new HRSGs have suffered damage.

Surveys show that about 70% of the HRSG failures are water-chemistry-related, compared to only about 30% in convention­al boilers. Reliable operation of HRSGs demands close attention to water chemistry both when the unit is operating and in lay-up (short term and long term). Also important: thorough inspections during annual outages to identify problems before failures occur and correction is possible.

The principal waterside fail­ure mechanisms are discussed in individual sections below. Guid­ance is provided to plant opera­tions personnel on where a given failure mechanism typically occurs and under what circum­stances—so you know where to look during the inspection. Next, you’ll see what the damage “looks like” to enable positive and rapid identification. How to correct the problem wraps up the coverage.

1. Flow-accelerated corrosion

Where failures occur. FAC can occur in almost all HRSG pip­ing that contains water (single phase) and water/steam mixtures (two phase). It is most common in and around the low-pressure (l-p) drum and l-p evaporator. Because water is an essential part of the FAC mechanism, it cannot occur in superheated steam lines.

What it looks like. Single-phase FAC can be described as an “orange peel” texture (Fig 1). A close-up view of the corroded area often appears as multiple comet-shaped divots in the metal. The damage starts and stops suddenly along the length of the piping but is often widespread around the circumference. It is more common in elbows, tees, reducers, or places where water or water/steam mixtures impinge on a metal surface.

Two-phase FAC occurs in pip­ing containing wet or saturat­ed steam. It has a “tiger stripe” appearance from magnetite being removed and deposited in dif­ferent areas by the water/steam mixture (Fig 2). In a cross-section under a metallograph, two-phase FAC often creates what has been described as “box” magnetite.

The mechanism. Several fac­tors combine to create FAC, which advances rapidly after it first begins. However, the reac­tion slows with time. Here’s why: The high solubility of Fe+2 in feedwater drives the mechanism.

Note that pure magnetite (Fe3O4) contains one Fe+2 for every Fe+3 in the structure. As the Fe+2 goes into solution during the corro­sion process, the concentration of the “protective” Fe+3 in the iron oxide deposit increases, thereby slowing the rate of metal deterio­ration.

  • Impact of alloys. Other oxides can substitute for Fe+3 in the deposit and have the same result. Most common is chrome oxide. The presence of even 0.1% chromium in the metal can make a significant difference between a pipe that has significant FAC and one that has none. In areas susceptible to FAC, alloys con­taining 1% chrome often are specified. Molybdenum has a similar effect.
  • Piping geometry and veloc­ity. While magnetite is ubiqui­tous in HRSGs, FAC is not. Obvi­ously, there is more to FAC than just the solubility of magnetite in feedwater. How quickly the Fe+2 migrates into solution deter­mines the rate at which FAC proceeds, and this is influenced by flow velocity and piping geom­etry. There is no minimum veloc­ity for FAC to occur; however, it increases with increasing veloc­ity. Where steam forms in the piping, conditions become more erosive, increasing the likelihood of FAC. Flow disruptions also are critical (Fig 3).
  • Water temperature is anoth­er important factor. As it increas­es, the solubility of magnetite increases. The solubility of sin­gle-phase FAC peaks in the 265F-300F range, two-phase between 300F and 390F—depending on which research you read. Above these temperature ranges, the characteristics of the oxide for­mation and the permeability of the base metal to hydrogen decrease the FAC rate. This does not mean that FAC cannot occur at higher temperatures, only that the rate is lower than it would be at lower temperatures. For more detail, see “FAC and cavitation. . . ,” COMBINED CYCLE Journal, Spring 2004; available at

Early detection. High iron con­centrations in the HRSG are one indication that corrosion, particu­larly FAC, may be occurring. The Fe+2 in solution can readily form magnetite, so simple qualitative tests—such as the Millipore fil­ter test—provide the information you need.

Prevention probably is an unrealistic goal when discussing FAC; minimization is more prac­tical, because the mechanisms that create the condition cannot be completely eliminated. The all-ferrous metallurgy of most HRSGs allows the use of chemi­cal treatment to minimize FAC. For single-phase systems, this consists of increasing the pH and eliminating the use of oxygen scavengers. Both minimize the formation of Fe+2 in the oxide layer. Typically, a pH of 9.6 is considered the minimum. The elimination of an oxygen scaven­ger alone generally is sufficient to maintain the oxidation-reduction potential (ORP) of the feedwater close to zero (slightly positive is thought to be best).

2. Corrosion fatigue

Where failures occur. Corrosion fatigue in HRSGs is most com­mon in and around economiz­ers, though it also is possible in evaporator sections. It is charac­terized by multiple failures; that is, if you have one, you’re likely to have several more, before all the affected tubing fails.

What it looks like. Corro­sion fatigue appears as a mul­titude of small oxide-filled cracks (Fig 4). The microscopic appearance of the cracks varies as the relative contributions of corrosion and stress vary. Where stresses are high, cracks will be relatively straight and nar­row—like thermal-fatigue cracks. Where corrosion is more preva­lent, the cracks will be branched, transgranular, and wider—often with areas of undercutting in the crack.

Chemical cleaning, particu­larly with inhibited hydrochlo­ric acid, has long been tied to an increase in corrosion-fatigue failures in conventional boilers. While the cleaning of HRSGs with inhibited hydrochloric acid is not a common practice, acidic species in the feedwater traced to contamination can produce the same results. Chloride and sulfate can concentrate at the bottom of the crack, accelerating the corrosion aspect of corrosion fatigue.

The mechanism. Corrosion fatigue, as the name implies, is a combination of two failure modes: a chemical or corrosion mode and a mechanical or fatigue mode. Every time tube metal tempera­tures cycle from cold to hot, stress­es occur at attachment welds or at tube penetrations in the headers. Stress cracks the protective iron oxide layer in the tubing, expos­ing fresh metal to the water. This then becomes a site for corrosion products to accumulate and con­centrate. Repeated cycles of stress under similar chemistry condi­tions act like a wedge holding the crack open.

Stresses that cause corrosion fatigue—like thermal fatigue—often are initiated by attach­ments, bends, or constraints. The difference is that the stresses are not sufficient to cause failure from thermal fatigue, but are sufficient to weaken the metal, making it susceptible to further corrosion.

The corrosion aspect of cor­rosion fatigue can come from several sources, including: pH, dissolved oxygen (particularly during startup), contamination, chemical cleaning solvent resid­uals (essentially hydrochloric acid), and treatment chemicals.

Early detection. Certain designs are more susceptible to corrosion fatigue than others. Because some of these designs are relatively new, participation in user group meetings, such as the HRSG User’s Group (, allows you to learn about designs that may have issues and what has been done to remedy them.

Prevention. The best way to address corrosion fatigue is to modify areas of high stress con­centration in a manner that less­ens or eliminates the potential for the problem. Also, keep in mind that monitoring and proper steam chemistry, particularly during startup, are critical to prevention.

3. Thermal fatigue

Where failures occur. Thermal-fatigue cracks are found in tight tube bends and tube-to-header welds. Stresses can be created during operation or be residual stresses remaining from manu­facture. Dissimilar metal welds and the heat-affected zone of the weld act as stress concentrators and are particularly susceptible.

What it looks like. The cracks are straight and narrow and often circumferential to the weld or stress concentrator (Fig 5). They are oxide-filled, similar to corrosion-fatigue cracks, but there is little evidence of branch­ing or pitting.

The mechanism. Differential thermal expansion between a tube and header, or at a weld, causes sufficient bending stress to crack the metal. The crack is filled quickly with oxidation prod­ucts that prevent it from clos­ing completely when the stress is relieved. The cycle continues until the crack is through-wall.

Early detection. Finite-ele­ment-analysis tools can be used to locate susceptible areas. Non­destructive examination (NDE) technologies—such as dye pene­trant and magnetic particle test­ing—also can be used to locate cracks. More extensive tempera­ture monitoring may indicate trouble areas. Additional ther­mocouples should be considered.

Prevention. Sudden changes in tube temperatures during startup often produce the stress­es capable of causing thermal fatigue. Operators can reduce the likelihood of thermal fatigue by implementing conservative startup procedures and limiting ramp rates. However, this often conflicts with environmental con­straints on the unit. The best pre­vention involves design changes that eliminate, or at least sim­plify, the structural interaction between components subject to transient temperature differen­tials. Specific issues to examine include these:

  • Tube flexibility between headers. This is critical to pre­venting thermal fatigue. Stresses are influenced by the arrange­ment of headers, tubes, and inter­connecting pipes in each section of the HRSG.
  • Interconnecting piping between the high-pressure evap­orator outlet headers and the h-p steam drum. As noted earlier, during startup, the front-row tubes absorb heat and estab­lish circulation more quickly than back rows, where condi­tions result in tube dry-out. This can create a “banana-shaped” evaporator. Interconnecting pip­ing must be flexible enough to accommodate this difference in tube temperatures.
  • Use of common headers with division walls (split headers) for tubes comprising different sec­tions of the condensate or steam process. Thermal stresses devel­oped in split headers may be tol­erable during normal operation, but can be extremely damaging during transient operation.
  • Tube-to-header welds. The type of fusion welds that attach tubes to headers and vessels can influence significantly the life of components subject to fatigue or creep conditions. Some experts claim that the fatigue life of full-penetration welds is at least 10 times greater than that of partial-penetration welds. Others dis­agree. See “Review basics of tube-to-header joints before writing specs,” 2005 Outage Handbook, COMBINED CYCLE Journal, Summer 2005; available at
  • Reheater and superheater materials and drains. Reheat­ers and superheaters for cycling HRSGs require special attention, because of their susceptibility to creep damage. Higher chromium materials, such as P91, and larg­er condensate drains often are recommended. However, beware that P91 presents its own chal­lenges, particularly in welding to carbon steel.

4. Under-deposit corrosion mechanisms

Under-deposit corrosion mecha­nisms include hydrogen damage, phosphate attack, and oxygen pitting. They are grouped togeth­er because they are all manifes­tations of the same problem—the concentration of an undesirable chemical species under a protec­tive deposit (Fig 6). The differ­ence lies in the chemical being concentrated.

Where failures occur. Under-deposit corrosion can occur in any tubing in contact with water, regardless of location. However, oxygen pitting tends to occur more often in economizers and l-p evaporators; phosphate attack is more common in high-temper­ature sections. High-heat areas are most susceptible to under-deposit corrosion, as are flow disruptions in the tubes—such as welds or bends.

Some corrosion occurs while the plant is offline, but more occurs as the unit is restarted and systems are fed oxygenated water. Even plants with auxiliary-heated deaerators can inadvertent­ly inject oxygen water if an under­size deaerator can’t keep up with the high demand for feedwater.

What it looks like. The appear­ance of this type of corrosion can vary depending on what is concentrating under the depos­it. By the time the leak occurs, the deposit that cre­ated the corrosion may be gone. But if there is under-deposit corrosion in one area, it usually can be found a few feet further down the tube or in an adjacent tube. Here’s how to identify the failure mechanism:

  • Oxygen pitting typically is broad and shallow and creates a pinhole leak at the thinnest point (Fig 7). If the tubercle is available for analy­sis you probably will find magnetite and hematite as well as chloride, which accel­erates oxygen pit­ting.
  • Hydrogen dam­age. The metal, when etched, has a very characteristic pat­tern of intergranular cracking or fissures created by a lack of pearlite at the grain boundaries (Fig 8). With finned tubes, the characteristic thick-lipped failure cannot occur.
  • Phosphate attack creates a very hard maricite (NaFePO4) deposit. It is also char­acteristic for the deposit to con­tain distinct black, red, and white areas. The deposits typically are undercut (similar to acid corro­sion) and there is a very sharp edge or boundary to the deposit on both sides, outside of which there is no corrosion. This is because the corrosion is extremely flow- and heat-flux-sensitive.

The mechanism. All under-deposit corrosion obviously begins with a deposit. The source of the deposit can be mill scale dating back to unit installation, hema­tite deposits from return con­densate, or contamination from water-treatment plant upsets. Regardless of the source, once the deposit is formed it creates an area underneath that is chem­ically different than the bulk water in the tube and this gener­ates the corrosion cell.

Oxygen pitting is very well understood because it occurs in so many situations both inside and outside the power industry. In oxygen pitting, there are instanc­es of high dissolved oxygen in the bulk water and low dissolved oxy­gen under the porous deposit. This creates an electrochemical poten­tial between the bulk water and the area underneath the deposit, thereby producing corrosion.

Dissolved oxygen can be absorbed into the water during a shutdown when water is left in the HRSG and the vents are open. Also, it can be entrained in the feedwater, particularly dur­ing startup when the deaerator or deaerating condenser do not function properly or are over­whelmed by the amount of aer­ated feedwater that is required. Heating the oxygenated water during startup accelerates the corrosion.

Hydrogen damage requires three things to occur: rapid for­mation of hydrogen gas (typi­cally from a low-pH excursion), a deposit, and heat. The low-pH condition results in iron and iron oxide corrosion and produces significant amounts of hydro­gen. Contamination, particular­ly chlorides, may create a local­ized low-pH condition even when the bulk water in the HRSG is alkaline.

The presence of a hard deposit prevents the nascent hydrogen from dissolving into the bulk water and forces the hydrogen to gravel through the metal along the pearlite grains. Here the car­bon in the pearlite reacts with the hydrogen to form methane, distorting the ferrite grains and creating voids or fissures. The heat both facilitates the diffu­sion of hydrogen through the metal and its reaction with the carbon.

For a one-time event, some damage may occur, but the metal will most likely only be weakened. Repeated instanc­es of hydrogen damage may be required before a failure occurs. Alternatively, the damage may attack an area already thinned by another under-deposit corro­sion mechanism and fail at that point alone.

Phosphate attack is the result of a reaction of magnetite with disodium phosphate, monosodi­um phosphate, or sodium hexam­etaphosphate to form maricite. Trisodium phosphate does not appear to react in this way and, to date, no phosphate attack episodes have been documented where only trisodium phosphate was used in an HRSG.

Phosphate attack can occur in several places—particularly at tube bends and areas that may be directly impacted by the use of duct burners. Poor circulation, particularly on startup, is suffi­cient to create phosphate attack in tubing.

Early detection. Traditionally, the thinning and corrosion cre­ated by under-deposit corrosion mechanisms could be found by ultrasonic thickness testing of bare tubes. For finned tubing, examination by borescope may provide indication of the presence of a deposit or its size, but current methods do not permit examina­tion of the depth of the deposits.

For hydrogen damage, some success has been reported with ultrasonic examination tech­niques on bare tubes, but here again, this would not work on finned tubes. Replacement of the entire section may be the only option when hydrogen embrittle­ment is found.

Preventing under-deposit cor­rosion is very simple: Get the waterside tube surface clean and keep it clean. Remove deposits and on-going corrosion stops, at least for a time. Chemical clean­ing is the best way to remove deposits. Preoperational clean­ing is still required to remove mill scale and other surface cor­rosion on many HRSGs. This step often is overlooked or delet­ed in the mad rush to get the unit operating. Not a good idea.

Similarly, using the proper feedwater and evaporator chem­istry with good monitoring can prevent many harmful deposits from forming in the first place. Once the deposits are formed, even good chemistry programs may not be sufficient to prevent under-deposit corrosion.

Many of the prevention strat­egies for oxygen pitting require capital investment. For example, you can’t expect the direct feed­ing of a chemical oxygen scav­enger to remove high levels of dissolved oxygen from your feed­water until the l-p drum begins generating steam. Dissolved-oxy­gen removal can be achieved with a startup vacuum deaerator or even a catalytic oxygen removal system. It could also be as simple as piping a steam source from another unit to provide steam for deaeration during startup. Good lay-up practices can also help sig­nificantly, by maintaining deaer­ated water in the HRSG during shutdown.

Phosphate attack can be mini­mized by eliminating the use of disodium phosphate or sodi­um hexametaphosphate (which becomes disodium phosphate in solution). Only trisodium phos­phate should be used for HRSG evaporator treatment. The other phosphate compounds often are used in vendor-supplied liquid phosphate products, so check the material safety data sheet or ask the vendor if you can use these. ccj oh

Idiosyncrasies of HRSG design, preoperational handling, and startup procedures can impact failure mechanisms

Waterside corrosion—as well as gas-side cor­rosion—sometimes can be traced back to the fabrication shop. One cause is poor stor­age procedures for tubes—outside in high-moisture locations, for example. In other instances, oxida­tion starts because metal surfaces are not passiv­ated or protected against the elements for shipment and storage at the plant site prior to installation.

However, corrosion that occurs prior to com­missioning generally is not a problem, provided it is recognized and proper cleaning and passivation of waterside surfaces are accomplished prior to first fire.
Many HRSG tube failures are associated with the thermal stress associated with unit cycling. Pure ther­mal fatigue is one of the common failure modes, but more often, the thermal stress adds to other water- and chemistry-related problems to create the corro­sion mechanisms experienced. These problems can manifest themselves within a few years and require extensive repairs. Common startup issues include thermal shock, circulation transients, and process upsets caused by equipment design and arrange­ment.

Thermal shock. During idle periods, condensate can collect in piping as well as in reheater and super­heater headers. Purge cycles on the gas turbine (GT) prior to firing may cool the superheater and reheater tubes further, thereby creating additional condensate. When firing begins, condensate that is not drained from the system contacts hot metal surfaces—such as the welds between tubes and the outlet header—and can initiate thermal fatigue cracks (see “Avoid desuperheater problems. . . ,” COMBINED CYCLE Journal, Fall 2004; available at

Circulation issues. HRSGs typically are designed with multiple rows of finned tubes attached to headers at the top and bottom in each economizer and evapo­rator section. This arrangement is in sharp contrast to the waterwall evaporator configuration in large power boilers. Sudden heating during an HRSG startup results in the first row of tubes in the gas path seeing a very different thermal profile than the last row of tubes in that bundle and results in circulation problems such as reverse and stagnant flow in some tubes.

The first rows instantaneously become risers and the last rows in the same evaporator section may act like downcomers—that is, until the unit is warmed up and the flow becomes uniform. Chemistry in the tubes during these periods is variable. If solids such as iron oxide, or dissolved non-volatile chemicals such as sodium hydroxide or sodium phosphate are present, under-deposit corrosion can occur. Atypical flow patterns that can occur while the GT is warm­ing up also can result in hot spots in some of the first generating sections.

Compounding these issues is the common prac­tice of using duct burners to generate additional steam; in some cases, almost double the HRSG’s unfired production rate. Duct burners also can create localized high-heat-flux and circulation issues that can result in under-deposit corrosion.

Design issues. In some HRSGs the deaerator is an integral part of the low-pressure (l-p) steam gen­erator. Oftentimes, the l-p drum is the only source of steam to the integral deaerator and, therefore, is not available when needed most—on unit startup. A high concentration of dissolved oxygen in the feedwater in the feedwater during startup leads to oxygen pitting and other under-deposit corrosion mechanisms, and accelerates corrosion fatigue.

The low-capital-cost mentality of many owners often excludes the installation of equipment that could improve cycling operation—such as a separate deaerator or an auxiliary boiler to provide steam to the integral deaerator during startup.