The fundamental purpose of a heat recovery steam generator (HRSG) is to extract the useful energy in waste heat, either from the exhaust of a gas turbine or reciprocating engine, or from the exhaust of some other combustion process in an industrial application. Here Robert Swanekamp focuses on HRSGs coupled to gas turbines and employed in cogeneration or small power production facilities.

When plant staffs were being recruited for cogeneration projects a decade or more ago, much of the focus was on the applicants’ experience with the specific model of gas turbine used on that site. Managers reasoned that their plant operators must intimately understand their ‘money machine’ if a project was to be successful.

To be sure, gas turbine performance was – and still is – a key to the success of a cogen project. But in recent years, many of the performance troubles that operators used to encounter with advanced gas turbine models have been resolved, or at least have become more manageable. In addition, gas turbines increasingly are covered by long-term service agreements, which shift the operation and maintenance (O&M) burden away from the plant staff and toward the turbine manufacturer. As a result, plant personnel today find themselves devoting less day-to-day attention to the sophisticated gas turbine, and more to the conventional steam cycle.

A major challenge for HRSG users is to find and repair a tube leak while it is still a minor problem. Approximately half of all such leaks occur in the economizer section, and are caused by fatigue, corrosion-fatigue, or freeze damage (Deltak LLC)
Click here to enlarge image

Contributing to this shift is the fact that heat-recovery steam generators (HRSGs) and their auxiliary steam systems are not so conventional any more, having increased in size and complexity. In fact, the three-pressure HRSG typically found in today’s plant should no longer be referred to as ‘a boiler.’ More properly, it is three boilers – a high-pressure (HP), an intermediate-pressure (IP) with a reheater, and a low-pressure (LP) boiler – each with its own unique purpose, materials of construction, O&M challenges, and water chemistry requirements. Realize too that these steam circuits typically are at higher temperatures and pressures, and must handle larger amounts of stored energy, than early HRSGs. Compounding the challenges is the fact that today’s HRSGs have substantially less access for inspection, maintenance and repair, compared both with older HRSG models and with typical fossil boilers.

Modern HRSGs have advanced over the past 15 years or so, in conjunction with the evolution of gas turbines. A variety of basic designs exist, depending on:

  • the specific characteristics of the waste heat – compositions of gas-turbine exhaust vary greatly in oxygen and water content, mass flow rate, turbulence and temperature
  • the purpose and duty requirements of the facility
  • the requirements to deliver steam for cogen applications or for extra steam production to produce more power during peak periods.

While steam-plant issues tend to be most vexing for utility-scale F- or G-class facilities operated in two-shifting service, operators and maintenance technicians at all cogen plants – regardless of size, shape and operating mode – should be on the watch for them. In fact, all steam systems, like all gas turbines, require a bit of TLC (tender loving care) if they are to provide long-term reliability in cogeneration service. Specific issues that most plants will face at some point in their service life include:

  • preventing and detecting HRSG tube leaks
  • maintaining optimum chemistry parameters throughout the water/steam cycle
  • combating dew-point corrosion in the cold end
  • reducing fatigue damage in the hot end.

Tracking tube leaks

At the annual HRSG User’s Group conference, the topic of tube leaks always seems to emerge as a challenging issue for users, manufacturing engineers, water-treatment specialists, and service professionals. Not only can tube leaks directly cause forced outages and lost revenue, they can also trigger subsequent damage to other components in the HRSG, which can be even more troublesome and more costly to repair than the initial leak.

In one conference presentation, Ron Meyer, Principal Engineer and Aftermarket Product Manager for Deltak LLC, explained that nearly half of all HRSG tube leaks occur in the economizer section, and that they typically are caused by fatigue, corrosion fatigue, or freeze damage. Another 25% or so occur in the superheater or reheater sections, often near the tube-to-header welds, Meyer has found. Thermal shock is the primary cause of these leaks, most often a result of overspray of the desuperheater, improper operation of the desuperheater control valve, or spray water leaking past a closed desuperheater block valve. A smaller percentage of tube leaks occur in the high-pressure (HP) evaporator. These leaks are typically at a mid-tube location, and are caused by deposits, which lead to corrosion or overheating. Steam specialists such as Meyer also find some leaks in the low-pressure (LP) evaporator, usually where there is a change in flow direction, making the area susceptible to flow-accelerated corrosion – a particularly dangerous failure mechanism that will be discussed below

For cogen operators, it’s a major challenge to find tube damage while it is still only a minor problem. For many years, operators of conventional boilers have used straight-wave ultrasonic testing (UT) to measure tube-wall thickness on a regular basis, and thereby monitor corrosion damage before it grows into a full-fledged tube leak. Unfortunately for cogen operators, the presence of finned tubing severely limits the use of straight-wave UT for tube-thickness measurement in HRSGs. A few unfinned areas of the HRSG that are susceptible to corrosion – such as tube bends or areas immediately downstream of tube-to-header weld joints in evaporators and economizers – can be examined by straight-wave UT, but only on the leading or trailing rows of those areas, where the tubes can be accessed. (Note: Several other UT techniques, more sophisticated than the straight-wave type, are being developed which may prove viable for the measurement of HRSG tube thickness.)

Because of the limitations of UT, most tube damage in HRSGs is not detected until it presents itself in an actual leak. Even then, detecting and pinpointing the location of that leak can be troublesome. When the HRSG is operating at load, vigilant operators may observe water draining from the casing, but HRSG specialists also recommend watching for mismatches between steam flow and feedwater flow, changes in the stack plume, and sudden difficulty in maintaining water-chemistry parameters. Other in-service monitoring techniques talked about by HRSG users include injecting helium into the drum while monitoring the exhaust stack with hand-held helium detectors, installing on-line acoustic-type steam-leak detectors like those applied in fossil boilers, and injecting radioactive sodium tracers as has been done in the nuclear industry.

In practice, existing leaks typically are located only by shutting down, cooling down, opening the unit for inspection, slowly re-pressurizing the suspect section in approximately 2-bar (25-psig) increments, and visually identifying the pinhole or crack that is the source of the water. Two cautions with this method:

  • If superheaters or reheaters are to be pressurized for this purpose, it is not unusual to also flood steam piping up to the isolation valves. For these piping sections that normally contain only steam, it’s important to pin or otherwise secure the spring and constant-support hangers, to prevent damage from the additional ‘dead load’ imposed by the weight of the water.
  • Sometimes compressed air, instead of water, is employed to pressurize these systems, while a directional sound gun is used to find the leak. In this case, air pressure must be limited to approximately 1 bar, because of the gas’s compressibility.

Once the location of a tube leak is pinpointed, Meyer recommends that the next step should be to find the location again. Applying a twist to the old carpenter’s adage about measuring twice and cutting once, Meyer says you should ‘locate twice, repair once.’ Most HRSG designs provide limited access for repair, forcing the maintenance team to remove and replace many good tubes to gain access to a failed one. So you don’t want to make the mistake of repairing the wrong section.

Keep current with chemistry counsel

The most important aspect of tube maintenance is not how to repair leaks, but how to prevent damage in the first place. That may also be a twist on another old adage – Benjamin Franklin’s 300-year-old aphorism that ‘an ounce of prevention is worth a pound of cure’ – but it is rarely practised in today’s cogen plant. So says Robert Anderson, principal of Competitive Power Resources Corp and a chairman of HRSG groups on both sides of the Atlantic (see box on p.58). He advises all of his clients operating steam plants to develop a formal tube-failure prevention programme – a systematic approach specific to each plant aimed at inhibiting pressure-part failures. For decades operators of conventional utility boilers have relied on such programmes to boost boiler reliability and reduce repair costs, but formalized programmes are rarely found in the HRSG community.

The basic tenets of a tube-failure prevention programme, according to Anderson, are: consistent collection, trending, and analysis of critical data; determination of the root cause when failures do occur; and timely, effective implementation of permanent repairs or modifications. Far too frequently, he says, HRSG operators faced with tube failures simply weld up the crack (destroying important metallurgical evidence), or rip out the damaged tubes, toss them in the dumpster, weld in some new ones made of the same material, and restart the plant. They end up with the same old materials in the same old designs being operated according to the same old procedures. Usually, this results in the same old tube failures, recurring again and again.

Flow-accelerated corrosion

One particular element that HRSG tube-failure prevention programs should address, Anderson emphasizes, is flow-accelerated corrosion (FAC). Also called erosion-corrosion, FAC is a dangerous and unseen failure mechanism that causes internal, localized thinning of pipe walls. It has led to many sudden, catastrophic failures of pipes, personnel injuries, and even deaths of power plant workers.

Flow-accelerated corrosion can cause sudden, catastrophicfailure of components – such as the rupture of this feedwater pipe – in steam plants of all shapes, sizes and technologies (Aptech Engineering Services Inc)
Click here to enlarge image

FAC is a function of fluid velocity, pH, temperature and oxygen content, all of which interact to create complicated, plant-specific riddles. In HRSGs, the LP evaporator section typically is most vulnerable because it operates in a temperature zone where magnetite becomes more soluble under certain fluid conditions. Anderson reports that some plants can reduce FAC by redesigning specific areas to change flow geometries and reduce localized fluid velocities. Other FAC-minimization tactics include replacing the vulnerable carbon-steel tube sections with components made of higher chrome alloys, and maintaining the feedwater pH between 9.2 and 9.6.

A parameter called the oxidation-reduction potential (ORP) also is key, explained Jim Witherow, executive chemist for Scientech LLC, at an HRSG User’s Group workshop conducted in November 2006. Most cogeneration steam systems built today contain all-ferrous components (no copper alloys), so Witherow recommends they operate in an oxidizing environment (defined as ORP equal to +100 mV). To achieve this environment, operators must avoid feeding reducing agents, which have been found to be a prime culprit in FAC occurring in HRSGs.

Unfortunately, accepting that advice requires a paradigm shift for many steam-plant veterans. For decades, reducing agents – also known as oxygen scavengers – have been fed to steam circuits in order to completely eliminate dissolved oxygen, and thereby prevent oxygen-pitting corrosion. So when chemistry authorities like Witherow recommend discontinuing their use, they also are saying it’s acceptable to allow some level of dissolved oxygen in the steam system.

This change in water-chemistry guidance is based on newer evidence, most of it compiled by the Electric Power Research Institute over the past decade, showing that operators need not drive dissolved oxygen all the way to zero. The EPRI studies show that levels of as high as 10 ppb are sufficiently low to control oxygen pitting, while the elimination of oxygen scavengers will reduce FAC.

Automated sampling

Another change in water chemistry guidance from years ago pertains to automated sampling and on-line monitoring systems. Dr Geoff Bignold, director of GJB Chemistry for Power Ltd asserts that the continuous, on-line systems available today can provide more consistent, more timely, and more accurate water chemistry data than ‘old school’ batch sampling. Bignold doesn’t supply such systems – he’s an independent consultant providing advice to power plants in the UK, Europe, North America, Asia and Australia in the field of chemistry for corrosion control. But he often helps clients develop a specification for their sample-conditioning system so that it will capture the required information to fully monitor steam-cycle chemistry. At a minimum, Bignold recommends that the on-line system should monitor the following parameters:

  • cation conductivity and sodium of condensate (to detect major condenser leaks)
  • specific conductivity of the makeup water in the makeup-water storage tank (to detect demineralizer upsets)
  • the pH of the HRSG drum (to maintain an alkaline environment to minimize corrosion.
  • cation conductivity of each HRSG drum (to determine the concentration of anionic contaminants)
  • cation conductivity of main steam (to evaluate anionic contaminants in steam to meet the steam-turbine manufacturer’s requirements).

Automating a plant’s water- and steam-sampling system – then integrating it with chemical feed systems – can provide more consistent, more timely, and more accurate water-chemistry data than batch sampling and manual testing (Waters Equipment Co)
Click here to enlarge image


How low can you go?

Dewpoint corrosion in the cold end is another common issue to which virtually all HRSG operators need to pay attention. At combined-cycle plants, the problem stems from the push for ever-higher thermal efficiency. At cogeneration facilities, the problem stems from the desire to export as much steam or hot water as possible.

Basic thermodynamics tells us that the lower the stack temperature, the greater the amount of energy that will be extracted from the gas-turbine exhaust. The problem is that if we allow stack temperature to drop below the acid dewpoint, moisture contained in the exhaust gas will condense on HRSG heat-transfer surfaces. That condensate typically is a highly corrosive sulphuric-acid mist.

Economizer tubes plugged with large rust flakes (left) and piles of rust on the HRSG floor (right) indicate a serious problem with cold-end corrosion (Chehalis Power Generating LLC)
Click here to enlarge image

Determining a specific stack’s precise acid dewpoint is challenging, because of variables such as the water vapour in the gas turbine’s exhaust and the amount of sulphur in the fuel. Note that gas turbines firing liquid fuels have the most trouble here, but even pipeline natural gas has some sulphur content. For a natural-gas-fired cogen plant, the acid dewpoint temperature may be in the range of 45ºC to 50ºC. Components most susceptible to the problem are found in the LP economizers and the condensate heaters. The inlet piping, headers, and tubes in these sections are at the lowest temperature and therefore are most vulnerable to dewpoint-corrosion attack.

To combat the problem, some owners specify high-grade alloys in these components. Others increase the temperature of the incoming feedwater, and the operating temperature of the LP economizer. This may require installation of a new – or retrofit of a better designed – recirculation system with its own pump and economizer-bypass control valve. Changes in operating procedures also may be needed, to enable heat-up of the LP economizer during each start-up and before each shut-down.

Drain it or damage it

Another potential damage mechanism in HRSGs of virtually any make, rating and vintage is referred to as ‘condensate quench.’ The problem is most common in superheater and reheater tubes where inadequate drain capability allows condensate to accumulate, causing serious flow restrictions and thermal stresses during start-ups and shut-downs.

Mike Pearson, president of J Michael Pearson & Associates Co (Georgetown, Ontario, Canada), was a pioneer in the research and understanding of this problem throughout the 1990s, working in much of that research alongside Bob Anderson. As Pearson explains, condensate forms in superheater and reheater tubes when the unit cools during a shutdown and also during the purge that’s conducted prior to the subsequent startup. This condensate formation occurs because the temperature of the air or exhaust gas cranking through the unit is below the saturation temperature of the steam residing in the tubes. The rate of condensate formation during purging can be particularly high, Pearson warns. And if a repeat purge is required because of a failed start, the amount of condensate can actually fill superheater panels a substantial portion of their height.

‘Condensate-quench’ damage occurs in superheater and reheater tube bundles when drain systems are incapable of removing the accumulated condensate prior to startup. The resulting elongated or buckled tubes often can be seen during HRSG inspection (Competitive Power Resources Corp)
Click here to enlarge image

If the condensate is not completely drained from all superheater and reheater tubes before steam flow is established during the subsequent startup – and at many combined-cycle and cogen plants it is not – two adverse consequences will result. First, condensate is ejected in large quantities into the outlet header and pipe manifold where it quench-cools the much hotter metal. The outlet header and manifold can be more than 200ºC above saturation temperature, the research of Pearson and Anderson has shown. The second consequence is that different tubes will have different average temperatures at different times. This results in differential thermal expansion stress between tubes. If the stresses are high enough, Pearson says, the cooled tubes will yield – or inelastically deform – stretching the tubes. When all tubes cool to ambient temperature at the next shutdown, the tubes that have been inelastically deformed will buckle because of their increased length.

If buckled tubes are present in an HRSG, they could be the result of just one start-up without proper draining, or they could be a sign that condensate-quench damage is occurring during every shutdown/startup cycle – which quickly will lead to thermal-fatigue failures. An increasing number of HRSG users report seeing signs of buckled or elongated superheater and reheater tubes during their annual inspections. That’s not surprising, because in a 2005 EPRI presentation to the HRSG User’s Group, researchers announced that thermal fatigue has now taken over as the leading cause of reported tube failures. Pearson adds that many of these failures, which manifest themselves as cracks in the tubes immediately adjacent to the toe of the attachment welds, occurred on HRSGs that had completed less than 10% of the starts that a cycling unit might be expected to perform during its service life.

What are users to do if buckled tubes are evident in their HRSG superheaters and reheaters? In their numerous published papers on the subject, Pearson and Anderson advise that a comprehensive response is needed, comprised of:

  • the installation of temporary thermocouples to evaluate the plant-specific transients
  • rewriting of start-up and shut-down procedures
  • re-engineering of the drain systems.

Resolving condensate quench is a complex undertaking, they warn, influenced by many variables – such as pre-start purge time, steam pressure of the HP system, internal superheater condensate formation rate, drain locations on headers, drain-pipe size, drain-pipe routing, method of condensate detection, type and number of drain valves, and control logic for automatic drain operation. ‘Failure to properly execute any one of these variables,’ Anderson concludes, ‘can seriously diminish the drain system’s effectiveness, and nullify the results of the entire retrofit project.’

Rob Swanekamp is Executive Director of the HRSG User’s Group, an international association with over 1500 members located in 50 countries and with headquarters in Bozeman, Montana, US.

Details on advanced ultrasonic testing techniques being developed are presented in the 2006 edition of the ‘HRSG Users Handbook,’ available through the HRSG User’s Group.

HRSG meetings coming this year

Robert W. Anderson will chair this year’s HRSG conference organized by the Fossil Power Committee of IMechE, scheduled for 30-31 October 2007, in Warwick, UK. For details, visit

For many years, Anderson also has served as chairman of the HRSG User’s Group, an international association open to all combined-cycle/cogeneration professionals. The Group’s next event will be a Technical Workshop, held 11-13 December, in conjunction with PennWell Corp’s Power-Gen International 2007 in New Orleans, Louisiana, US. For details visit or