HRSGs for small combined-cycle and cogen plants

Size matters, or so it seems. Take heat – recovery steam generators , for example. Mention “HRSG” and most people in the power business think of the triplepressure units in service at large combined-cycle plants. Articles in the trade press focus on these units, as do papers at major industry conferences.

That’s interesting, because during the recent building boom associated with gas-fired generating facilities, so many HRSGs were built by two or three manufacturers for service behind only two or three gas-turbine (GT) models one could almost say they were “mass produced.” Not much to talk about under that scenario.

Drop down in size to the HRSGs serving in small combined-cycle and cogeneration plants—those powered by GTs producing up to about 60 MW—and it’s another world in terms of design variation. The reason is simple: Large high-pressure/high-temperature reheat HRSGs burning natural gas—and possibly some distillate—are optimized for maximum electricity production and, in most cases, base-load service.

Small HRSGs often are unique—customdesigned to (1) serve a specific process, (2) satisfy a specific operating requirement, (3) burn a waste or intermittent fuel, etc. Some engineers mistakenly believe that HRSGs for the smallest GTs in industrial energy systems (assumption here is machines in the range of 1 to 5 MW) are simply “light-duty, commercial- type boilers.”

That’s not accurate. Public utilities and industrial companies that buy many small HRSGs specify watertube boilers built to Section I of the ASME Boiler & Pressure Vessel Code that are as robust as much larger units.

Perhaps such thinking prevails because many specifiers and purchasers of small HRSGs are contractors or engineering firms that do not have boiler specialists on staff. To assist those companies in the preparation of meaningful specifications, Bob Krowech, president, HRST Inc, Eden Prairie, Minn, a firm specializing in heat-recovery boilers, provides guidance in the accompanying sidebar.

The small-HRSG market is served by several firms offering conventional drum-type and oncethrough steam generators (OTSGs). The leaders in the supply of drum-type boilers for GTs rated up to 60 MW are Rentech Boiler Systems Inc, Abilene, Tex, and Express Integrated Technologies LLC (formerly ATS/Express LLC), Tulsa, Okla. Rentech specializes in the bottom half of the size range and Express on the larger units and on special applications throughout the sector.

CMI EPTI LLC, Erie, Pa, the firm resulting from the purchase by Belgium’s CMI of Erie Power Technologies (formerly Aalborg) also competes here, interested primarily in naturalcirculation HRSGs serving GTs larger than 20 MW. Rob Dueck of Foster Wheeler North America Corp’s Canadian operation (, 905-688-4587) says the LM6000 marketed by GE Energy, Atlanta, which can be designed to produce from 40 to about 50 MW depending on enhancements, is at the lower end of its HRSG product line. This is true also for Vogt Power International Inc, Louisville. However, Vogt is active in retrofit and upgrade projects for HRSGs of all types and sizes, as the second sidebar reflects.

There are several other boiler manufacturers capable of building drum-type HRSGs for GTs in the size range discussed here, but the editors received no evidence of current interest. OTSGs, covered in detail later in this report, are the specialty of Innovative Steam Technologies (IST), Cambridge, Ont, Canada, and Alstom Energy Recovery Systems Ltd, London.

Tight specs key to reliable operation, top performance
“It’s a Catch-22 situation,” says Bob Krowech,president, HRST Inc, Eden Prairie, Minn,referring to the preparation of specificationsfor small heat-recovery steam generators (HRSGs).“While the typical purchaser generally has lessexperience than purchasers of large, triple-pressureunits, most cogen applications for small HRSGs areunique and may actually require a broader experiencebase to ensure the development of a satisfactoryspec. Additionally, bidders may includesuppliers that have limited HRSG experience,making thorough specifications particularlyimportant.

“However,” Krowech reminds, “performancespecifications that require adequacyfor the intended service do little to preventproblems. When problems occur, there isthe potential for disagreement about whetherthe owner operated the unit within theintended parameters. It is far better to avoidthe problems in the first place than to try toenforce warranties.”

Krowech, who founded HRST to specialize in thedesign, troubleshooting, and inspection of heatrecoverysteam generators, also suggests carefuldue diligence of alternative suppliers. “You want tobe sure the vendor selected has experience that’srelevant to your project,” he says. “For example,if duct burners will burn waste gas, does the candidatesupplier have a successful unit to refer youto? Keep in mind that tube and fin spacings, as wellas other design parameters, may vary significantlydepending on fuel physical and chemical characteristics.

”Here are nine key items that Krowech suggestsshould be included in your next spec:

  • Tube spacing and fin density. The lane betweenfin tips should be adequate to accommodate periodiccleaning and routine inspection. A small incrementin tube spacing can profoundly impact yourability to clean the boiler. This is critical ifthe unit is oil capable or if an SCR (selectivecatalytic reduction system for reducingemissions of nitrogen oxides) is installed. AllHRSGs must be cleaned eventually.
  • Access lanes. (1) Lowest first cost favorslarge tube bundles with few access lanes.Specify minimum lane width: Consider thelane size needed for a sky climber, if applicable.(2) Also specify maximum bundle size(number of tube rows) that can be cleaned,repaired, and inspected.
  • Steam drums traditionally have been sized basedon retention time. Experience indicates that this isacceptable for single-pressure boilers, as well asfor high-pressure drums in multi-pressure HRSGs.However, it does not work for intermediate-pressure(i-p) and low-pressure drums in multi-pressuresystems. There are many HRSGs in service that aredifficult to start because of level-control problemsin the i-p steam drum. All steam drums should be sized for both retention time and surge volume.
  • Flow-accelerated corrosion. This corrosionmechanism often goes unnoticed until damage isdone because it occurs on the waterside in placesgenerally not accessible for inspection. The rootcause of FAC is a combination of materials selection,waterside velocity (and flow instability in theevaporator section), and water chemistry. TheHRSG supplier is responsible for waterside velocityand the materials used.
  • Liner and insulation. (1) The exhaust flow fromsome small turbines—for example, the LM6000 andLM2500—is very turbulent and the duct liner andinsulation are subject to severe buffeting. There aremany designs and combinations of stud spacing,stud size, and liner thickness that work, and manyother designs that are forever troublesome. Linersthinner than 12 gauge (the smaller the number thethicker the material) and pins less than ½ in. diametershould be viewed with suspicion. Ultimately,it is the overall design combination that works ordoesn’t.

    (2) Firing ducts are subject to local temperatureshigher than the nominal firing temperature. The firingducts on small HRSGs are more susceptible to localoverheating than those on large units, because thediameter-to-length ratio is smaller. The combinationof liner thickness and material, and stud spacing,determine the success of the firing duct liner. Besure to get the bidder’s design temperature for thefiring duct. There should be a substantial marginbetween the firing temperature and the design temperature.

  • Superheater-tube metal temperature. (1) Thetube metal temperature used for designing superheatertubes should have a specified margin overthe calculated tube metal temperature. In a perfectworld, the margin could be zero.

    (2) Tube skin thermocouples can be used tomonitor tube temperature. Installation of these thermocouplesis not trivial. Be sure that the installationmethod is a proven one that gives meaningful output.

  • Waterside inspection and maintenance. Includeaccess for borescoping and connections for chemicalcleaning in the specification. The cost of includingthese items in the original scope of supply ismuch less than the cost of retrofit.
  • Cycling service. (1) The bidder should includethe allowable ramp rates for startup and shutdown.Additionally, the issues of economizer thermal shockand economizer steaming should be addressed.Make sure there is a plan for a cold start during thecoldest time of the year. Dry lay-up is a perfectlygood way to prevent freezing. But keep in mind thatfilling a cold boiler in freezing conditions requiresexpertise.

    (2) Consider a stack damper if cold-weather layupis likely, even if the HRSG is indoors. The naturaldraft through a warm boiler can consume a largeamount of heat.

  • Instrumentation. Intermediate gas- and watersidetemperatures (that is, temperatures at otherthan the inlet and outlet) are not needed on a dailybasis. But they are very useful, even necessary,for performance analysis and troubleshooting. At aminimum, specify test ports and thermowells—andremember that gas-side thermocouples should haveradiation shields to ensure meaningful output forcalculations.

Drum-type HRSGs

Bob Gdaniec, CMI EPTI’s engineering manager (, 814-897-7101), has been involved in the design of HRSGs for about two decades. He sees more interest in small HRSGs today than at any other time in the last five years. As a designer, he encourages customers to be realistic about the boiler’s intended duty cycle before engineering begins.

Owners always are cost-sensitive, he says, so his challenge is to develop a cost-effective design to accommodate the duty cycle—which makes understanding the needs very critical. For example, if the unit is intended for base-load service, rolled tubeto- drum joints are satisfactory in this size range. But if regular cycling is planned, rolled and welded joints, or an all-welded construction, should be considered to minimize maintenance and maximize availability. It is very difficult to change the design after the HRSG is installed.

As a starting point for specifiers, Gdaniec suggests a single-pressure HRSG to serve GTs rated up to about 20-25 MW. “You can comfortably produce steam in the 650F to 700F range with such a boiler,” he adds. Dual-pressure HRSGs probably make most sense when the GT is in the LM6000 class of machines.

Gdaniec says it is common for HRSG owners to push thermal output beyond what GT exhaust heat alone can produce, especially in the small-gasturbine/ HRSG size range. He has designed several boilers for LM2500s that can nearly double the unfired steam capacity with a duct burner. Adding to system complexity, many of these supplementary firing systems often involve multiple fuels in addition to natural gas—such as distillate oil, kerosene, and/or refinery gas.

CMI EPTI uses global sourcing to help owners hold down costs. Today, the company generally sources tubes in Europe and does finning, as required, in Korea. Headers and drums also are made in Korea. Right-sized modules and tube bundles are shipped to the job site.

Structural steel and casings/ductwork also are sourced globally, with attention paid to the proximity of suppliers to the final destination. Delivered cost often dictates fabrication in-country, near the job site. Valves, instruments, fuel skids, duct burners, and most boiler auxiliaries come from USbased specialty vendors. Note that CMI EPTI has no US-based manufacturing facilities.

Gdaniec offers these “pearls” for owners considering the purchase of a small HRSG:

  • Specify spargers or heating coils, and stack dampers, to permit hot standby and enable quick starts. “In cycling service,” he says, “an ounce of prevention can go a long way.”
  • Be conservative on materials for superheaters to ensure long life and reliable service. Anticipate the upset case and plan accordingly.
  • Encourage a generous length for the combustion chamber, this to protect against flame impingement on tubes in the first few rows. Predicted flame lengths should be checked across the range of operation.
  • Support the idea of no metal liner in ductwork downstream of the burner. Experience indicates that the ceramic fiber modules available today perform well and avoid the inevitable headache of liner maintenance. For small HRSGs with a high degree of firing, this can be a maintenance advantage.

Express designs and builds HRSGs for GTs in the range of 5 to 150 MW, but concentrates on medium- size to small projects with a high degree of custom engineering—such as a heat-recovery system designed to maximize thermal output while minimizing emissions—according to VP Philip Childers (, 918-622-1420).

“Some manufacturers ‘mass-produce’ quasi-standardized, commercial-grade boilers for small GTs,” he continues. “Express provides HRSGs requiring project-specific design considerations that allow us to offer value-added engineering for more robust, industrial-grade systems.”

To illustrate the company’s strengths, Childers points to the HRSG it designed to accommodate the innovative Sconox process selected by the City of Redding’s (Calif) electric utility to simultaneously remove NOx and CO (Fig 1). The boiler sits behind one of the first GTX 100 GTs installed in the US. Express shared in the Powerplant Award presented to Redding Electric Utility last April (access COMBINED CYCLE Journal, First Quarter 2004, at psimedia. info/ccjarchives.htm for more detail).

“Express provided most of the medium-size HRSGs that were installed in California during the peak, when other manufacturers were concentrating on ‘cookie-cutter’ F-class units,” states Childers. The HRSG for the LM6000 at Modesto Irrigation District’s Woodland Unit II is supplementary fired to increase steam production to 250% of unfired capacity, all while meeting ultra-low NOx, CO, and ammonia- slip requirements. The HRSGs for the LM6000s at the City of Santa Clara’s Silicon Valley Project have stringent emissions performance requirements like those for Modesto; however, the duct burners at Silicon Valley are not fired quite as hard.

Another example: The company currently is designing boilers to sit behind three Mars 90 GTs (Solar Turbines Inc, San Diego) installed a few years ago at a West Coast wastewater treatment plant. The original HRSGs are being replaced. Those standardized, package- grade boilers were incapable of accommodating the special cleaning requirements posed by the deposition of siloxane compounds found in GT exhaust gas when burning digester gas. There will be more coverage of this project when the new boilers are installed early next year.

Rentech specializes in natural-circulation steam and hot-water HRSGs designed to Section I of the ASME Code for GTs up to 25 MW, according to Senior Sales Engineer Kevin Slepicka (kslepicka, 402-474-4242).

The young company’s largest unit in terms of exhaust-gas flow is the HRSG supplied for the Corona combined cycle profiled elsewhere in this issue. All boilers are made in Rentech’s large modern shops located right off I-20 in Abilene, Tex. Slepicka, who has been involved in HRSG design for more than a decade, says US manufacture is a competitive advantage when customers want to maintain tight control of their projects. Rentech’s shops are easy to access and the HRSGs in the size range offered all are transportable by truck or rail in modules of convenient size.

A typical small HRSG supplied by Rentech has ceramic fiber insulation sandwiched between an outer carbon steel casing and inner stainless liner. The evaporator section, consisting of carbon steel tubes and downcomers, is integral with the steam and mud drums as shown in Fig 2.

Supplementary firing sometimes is installed to boost steam flow as it is in Fig 3. The boiler shown was designed to recover heat from a 1.5-MW GT installed at a West Coast college. The unit makes 10,000 lb/hr of steam on exhaust gas alone, double that with the duct burner in operation. The majority of Rentech’s jobs are for saturated-steam service, like this one.

When superheated steam is required—for example in combined-cycle service— steam pressures go to more than 800 psig, temperatures to well over 900F. T22 tubes are used in the superheater for such applications.

Boiler designers must have accurate information on expected ambient temperature and electrical and thermal outputs over a typical operating year to do the best job for the owner. Slepicka says that the need for supplemental firing can have a tremendous impact on boiler design. For example, under certain operating conditions there is the possibility that the flue gas may not have enough oxygen to support firing of the duct burner to the extent needed to meet the thermal requirement.

Don’t forget the HRSGs when you upgrade gas turbines

Robert Threlkeld, who manages Tenaska Inc’s Central Alabama and Lindsay Hill combined cycle facilities, advises elsewhere in this issue, “Refuse to live with problems. If something does not work properly, correct it or find a better way.”

It’s one thing to identify a problem, another to solve it. Sometimes the expertise exists on staff, sometimes not. And even when a viable solution can be identified in-house, the staff resources may not be available to implement it. Then, too, when it gets down to the details involved in engineering a solution, plant personnel might not have answers to all the questions that arise—it’s not their job.

Consider the problem of a capacity shortfall at an cogen plant equipped with a GELM5000 gas turbine (GT). Upgrade packages exist to squeeze more power from these units and to improve their reliability at the same time. The upgrade essentially creates an LM6000 from the older unit. Assume that the upgrade is possible electrically, because the generator, associated switchgear, and transformer can accommodate the power increase.

With a green light to this point owners often think that the project is a “slam dunk”—when budgeted, of course. “Not necessarily,” says Jeff Daiber, director of aftermarket services for Vogt Power International, Louisville (, 502-899-4515).“It depends on the design conservatism built into the heat-recovery steam generator (HRSG). We have seen several instances where the upgraded GT’s higher exhaust-gas flow rate and temperature would push the HRSG out of compliance with Section I of the ASME Boiler and Pressure Vessel Code if designed today using the standards as they originally existed. Any pressure part that requires an “R” stamp must “include calculations verifying the new service conditions.”

Les Elwonger(,502-899-4506), the GT expert on Daiber’s aftermarket “swat “team, says that you can expect an upgrade from LM5000 to LM6000to increase exhaust temperature by about 15 deg F and the exhaust flow rate by a few percent—together boosting exhaust gas energy by about 5%. Result is higher GT back pressure and increased steam production.

“Any time you change the heat source,” Daiber continues, “it is good engineering practice to check boiler design pressures and temperatures, pressure parts, and the relieving capacity of the safety valves.” Failure to do so could put the owner at risk if an accident were to occur. The existing boiler and machinery insurance policy might not cover the upgrade. If the heat source is changed, you may need to re-reate the boiler and, according to the National Board Inspection Code (NBIC), this requires an “alteration” (more detail below).

Elwonger says that for the LM5000-to-LM6000upgrade, Code-required calculations may indicate excessive steam temperature and find that the attemperator installed is unable to handle the additional thermal load. In some cases, you may have to add a coil upstream and/or downstream ofthe super heater, if installed. Or perhaps retube the super heater with a more heat-tolerant material.

Safety and relief valves on the boiler and in the accompanying steam system also must be checked to verify relieving capacity at the new conditions.“Don’t be surprised,” adds Daiber, “if they don’t make the grade. Review the adequacy of steam piping external to the boiler, too.

“Sometimes contractors hired by the plant don’tanticipate the analytical effort required with respect to the HRSGs when upgrading a GT, and regardit as a nuisance or ‘tea kettle’,” continues Daiber.“But it often proves to be a very positive exercise for more than just safety reasons. In many instances where a cogen system has been in service for more than five years, or so, operating conditions have changed and a cycle analysis can point the way to improved efficiency.”

Trying to squeeze more power from an existing unit is not the only reason for rerunning boiler calculations. Another, Elwonger says, is to gauge the impact on the HRSG of normal GT degradation over time. “Remember,” he adds, “that the exhaust temperature of aero derivative machines goes up as the engine degrades.” The opposite is true for frame machines.

Boiler turndown (as a percentage of rated steaming capacity) beyond the 20% or so typically specified by purchasers is relatively common today, says Daiber. “We see some owner/operators wanting turndowns double those designed for. Don’t assume that this can be done on a regular basis without adversely impacting reliability, availability, and efficiency—and possibly safety.”

Circulation is an issue for sure, Daiber adds, and suggests you also re-evaluate the performance of the emissions control system at the lowest load you plan to operate at. “Your plant could wind up out of compliance at 40% turndown if it is designed for20%,” he states.

Daiber suggests that while most boiler owner/operators are familiar with the ASME Code, few are expert in the changing nature of its content. Also, he finds some owner/operators are not knowledgeable about the NBIC, which governs modifications and repairs after commissioning. It is published every third year (current edition is 2004, expected as the COMBINED CYCLE Journal went to press) and updated annually like the ASME Code. Focus of the NBIC is on safety.

Perhaps of greatest importance to boiler owner/operators is that any HRSG repairs and/or alterations are governed by the NBIC. For example, plans for repairs—such as the replacement of any pressure part in kind—requires review by a professional engineer and signoff by an authorized National Board inspector.

Alterations—including such things as an increase in the maximum allowable working pressure or temperature, or increasing the amount of heating surface such that an increase in relieving capacity is required—call for revised boiler calculations, re-rating in accordance with the original Code edition of construction, pressure testing, and, perhaps, adherence to special methods for welding and post-weld heat treatment.

To illustrate how dramatically temperatures and flows can vary under different operating regimes, Slepicka offers the following example: A 5.3-MW Solar Taurus 60 produces about 165,000 lb/hr of 950F exhaust gas when operating at full load. At 50% of the full-load rating you get about 100,000 lb/ hr of 1100F exhaust. The question the owner must ask: Do I need full steam capacity when the turbine is operating at half load?

Let’s assume that the answer is “no” and supplementary firing is not required. For a boiler producing saturated steam, half-load operation demands no special consideration. However, for a HRSG designed to produce superheated steam, the gas flow and temperature at reduced load might dictate a change in superheater arrangement, a different tube material, and revised thinking with respect to the desuperheater.

Supplementary firing. If you can’t achieve the necessary thermal output when the GT is reduced to half load, a duct burner must be installed. This is where the boiler designer earns his pay. The variable that really makes a difference in boiler design is the maximum gas temperature downstream of the duct burner.

Rentech designs its units for high reliability and long life, says Slepicka, meaning it wants to hold the gas temperature at the burner exit to 1600F in normal operation. At this temperature, the company’s standard sandwich casing of stainless liner, ceramic fiber insulation and carbon steel exterior noted earlier works well and the owner benefits from the cost-effective design. But 1600F is not always possible because of the thermal demand.

Consider the following: An unfired HRSG on the back end of a Taurus 60 produces about 25,000 lb/hr of saturated steam; with supplemental firing and a final gas temperature of 1600F, 60,000 lb/hr is possible. But these numbers are for STP conditions (59F ambient at sea level). If the outside air goes to 100F and there’s no turbine inlet cooling system installed to keep exhaust flow constant over the ambient temperature range, exhaust-gas flow will drop. Thus maintaining 1600F at the burner outlet means that steam output will probably drop by a few thousand pounds per hour.

You can push the burner to make up the shortfall, but this is not a recommended operating strategy. Rentech has a safety shutoff when the burner exit temperature increases to the 1700F-1750F range to protect firing chamber integrity, evaporator tubes and fins, etc.

Of course you can get more steam from a HRSG downstream of a Taurus 60, but the design would change dramatically. Let’s say you need 100,000 lb/hr. This could be achieved by designing for a burner exit temperature in excess of 2000F. To accommodate such a high firing temperature, Rentech would offer a design that includes 100% membrane waterwall construction. The resulting unit looks very similar to a conventional fired packaged boiler. The O-type boiler in Fig 4 is an example. It sits behind a Taurus 60 at a military base and produces 80,000 lb/hr of 250-psig saturated steam.

Fresh-air firing. Proper sizing of the duct burner is important to ensure the desired thermal output under all operating conditions. The designer’s greatest challenge, perhaps, is to produce rated steam output when the GT is not operating. This requirement is not unusual for cogen systems installed in process plants where a drop in steam pressure and/ or flow can have severe financial consequences.

A rule of thumb: A “standard” duct burner used in supplementary-firing duty is only capable of producing about half the rated steam flow when the GT is out of service. Thus, if rated thermal output were critical, you need a much larger burner than normally would be installed and it would operate in the so-called “fresh-air-firing” mode when the GT is shut down.

However, when the GT is in service, a standard 10:1 turndown burner cannot provide the operating flexibility normally required (its maximum firing level with GT exhaust gas available would be 50% of the burner’s rated output). A burner and the control system required to provide a 10:1 turndown with the GT at full output must have a wider operating range and it, and the companion control system, cost more than a standard supplementary firing system.

Slepicka points out that designers can build boilers to accommodate virtually any operating conditions specified by the owner. In the end, it’s just a matter of cost to achieve the flexibility desired. Consider the situation where a transfer to fresh-air firing must occur seamlessly—without a decrease in steam pressure and output.

The most common approach is to install forced-draft fans and have them start on a GT trip. However, the challenge here is to get sufficient fresh air into the furnace quickly enough to keep the burner operating in accordance with NFPA (National Fire Protection Association) and other safety codes. A fan failure on start would trigger a shutdown, and a furnace purge would be required before restart. It is unlikely that a critical process could tolerate such a transient.

Another option is to install louvers upstream of the duct burner and an induced-draft fan just ahead of the stack. The fan runs continuously—an obvious operating cost. With this arrangement, when the GT trips, fast-acting air inlet louvers open automatically and burner output is ramped up quickly.

Emissions control. Keep in mind that selection of the larger burner for fresh-air firing has implications for the emissions control system as well. It will require more CO and NOx catalyst than normally would be installed if the exhaust heat from a natural-gas-fired GT equipped with a state-ofthe- art low-NOx combustor were producing half the thermal output.

Slepicka says that integrating emissions control into HRSG design and achieving the owner’s goals for both CO/NOx and thermal energy is a difficult job, one characterized by tradeoffs with almost every decision. For HRSGs in process plants this is particularly challenging given the variety of fuels and operating conditions that often characterize that industry sector.

On occasion, it is impossible to install catalyst in a “conventional” boiler layout and achieve all design goals. The optimum solution sometimes involves splitting the evaporator section, as shown in Fig 5. This allows positioning of the catalyst in the gas stream for maximum reactivity, thereby reducing the amount of catalyst required.

Once-through steam generators

OTSGs, the acronym for once-through steam generators, are an alternative to drum-type HRSGs for GTs in sizes from about 5 to 60 MW. An OTSG, in its simplest form, is a continuous-tube heat exchanger in which preheating, evaporation, and superheating of the feedwater takes place in series (Fig 6).

An actual OTSG (Fig 7) has many tubes mounted in parallel and joined by headers—thereby providing a common inlet for feedwater and a common outlet for steam. Water is forced through the tubes by a boiler-feed pump, entering the OTSG at the cold end (top). It changes phase along the circuit and exits as steam at the hot end (bottom) of the unit. Exhaust gas flows in a direction opposite to that of water and steam, as shown.

OTSGs, unlike drum-type HRSGs, do not have defined economizer, evaporator, and superheater sections. Nor do they have steam drums, mud drums, or blowdown systems. The point at which the steam/water interface exists is free to move through the tube bank, depending on heat input, mass flow rate, and water pressure.

“The OTSG has a single point of control—the feedwater control valve,” says Ryan Tangney of Innovative Steam Technologies. Tangney (519-740- 0757, x-261; provides engineering support for the North American market.

When operating at full load, the feedwater flow rate is regulated using a feedforward and feedback algorithm programmed into the plant’s distributed control system. The DCS senses any changes to the GT exhaust or steam conditions and quickly adjusts the feedwater rate to produce the desired outlet temperature. Such a control scheme, says Tangney, “permits regulation of steam temperature to within 5 deg F when operating at full load—without the use of a desuperheater.”

Selecting between drum-type and oncethrough HRSGs is relatively simple in some cases. For example, if a capital budget has been prepared based on a very competitive drum-type thermal system, it’s unlikely that an OTSG bid will meet the owner’s expected price. Once-through boilers require demineralized water and alloy tubes; lower-cost alternatives are available for drum-type units.

However, if a cogen system is being retrofit into an existing building, variables such a footprint and floor loadings may be sufficiently important to make an OTSG the clear-cut choice.

Although OTSGs traditionally have been more expensive than drum-type boilers on a first-cost basis, Tangney urges that selection between the two be made based on life-cycle cost. He believes that the OTSG is less expensive to install and that it has efficiency and operational benefits compared to conventional steam generators that over the design lifetime give the once-through unit a competitive advantage.

Here are some notes on the differences between OTSGs and drum-type HRSGs to help guide your decision-making:

  • Water treatment. OTSGs require demineralized water to preclude the possibility of solids deposition in the tube bundle. IST says total dissolved solids (TDS) must be maintained at less than 50 ppb; cation conductivity to less than 0.25 ?S/cm. The boiler-water and makeup treatment systems for drum-type boilers typically installed in cogen and combined-cycle systems with GTs rated 60 MW and below are less costly to own and operate.

    However, if the GT features water injection for power augmentation, a demineralizer already is required for the plant. The cost of expanding this system to also serve the boiler is significantly less than the financial penalty that would be assessed were the demineralizer installed to serve the boiler alone.

    No blowdown is required for OTSGs, reducing makeup requirements compared to drum-type units by about 95%. No blowdown also means there’s no penalty assessed for its treatment prior to discharge, if required by regulations, and for the thermal loss associated with dumping hot water to the sewer on a continuous, or nearly continuous, basis.

  • Tube materials. The modest steam temperatures associated with most small HRSGs permit the use of relatively inexpensive carbon steel boiler tubes. OTSGs, by contrast, generally rely on premium high-nickel steel tubing (Incoloy— or, generically, Alloy—800 or 825. This, of course, impacts first cost, which is of primary concern to the majority of project developers. The material also can adversely impact maintenance costs because of the sophisticated welding process (TIG, for tungsten inert gas, and special orbital welding gear) required in the unlikely event of a tube leak.

An advantage of the high-strength, thin-wall alloy tubing, says Tangney, is that it allows the OTSG to start up, shut down, and respond to load changes rapidly without exceeding material stress limits.

It also enables the OTSG to run dry, unaffected by hot GT exhaust. This eliminates the need for a damper and bypass stack were they required. A point to keep in mind when weighing the value of operating the GT without recovering exhaust heat: If an SCR is installed for NOx control, the catalyst must be able to operate at the higher temperature—or the exhaust temperature must be reduced to avoid premature catalyst degradation.

Cold-end corrosion resistance is another attribute of Alloy 825. One manufacturer uses it together with stainless steel fins in the “economizer” portion of its OTSGs and can handle inlet feedwater temperatures down to 60F without preheat. This lowers the stack temperature and improves cycle performance.

  • Footprint. OTSGs configured for cogen or combined-cycle operation generally have tubes in a horizontal (refer again to Fig 7), or helical arrangement, allowing a smaller footprint than most traditional HRSG designs. Addition of an SCR and CO catalyst does not add significantly to the footprint.
  • Modular or single-component construction and remote operation are two advantages claimed by OTSG manufacturers, but both sometimes can be accomplished as well, or nearly so, with conventional HRSGs. As a point of comparison, Tangney says IST can complete the installation of an OTSG for an LM6000 within four to six weeks.
  • The minimal instrumentation required on OTSGs and their dry lay-up requirement sometimes lead suppliers to suggest that a building is not required for these units in cold climates as it is for drum-type HRSGs. Perhaps.

    Tangney suggests keeping an open mind on outdoor installation. He says IST is installing a unit in Alaska where the design minimum ambient temperature is -61F, adding that about half of the company’s installations are in locations where freezing is a serious concern and none have building enclosures.

Two manufacturers offer OTSGs. IST is the dominant supplier of once-through boilers for GTs 60 MW and less. Tangney says the company has supplied 99 units worldwide—almost one third of them to owners of LM6000s. In round numbers, about 15% of the orders were for machines built by Solar; another 10% each serve on LM2500s and Hitachi Ltd’s (Tokyo) H-25s. IST’s once-through boilers are designed to Section I of the ASME Boiler & Pressure Vessel Code.

Alstom Energy Recovery Systems Ltd, London, has developed and commercialized, over the last five years, a patented OTSG product line for the oil and gas industry sector known as CiBas (pronounced sea bass)—for concentric internal bypass and silencer. The company has five packaged designs to serve GTs from 5 to 50 MW, says Dave Normandale, business development manager (+44 208 799 3244, com. CiBAS boilers, designed to Section I of the ASME Code, are suitable for both offshore and onshore applications.

The CiBAS design is significantly different from the IST offering. For example, the Alstom unit is circular in configuration, pressure parts are helically coiled, and variety of tube materials is offered from carbon steel through Incoloy 800H depending on operating conditions.

The circular design was selected because it requires a minimum amount of steel. For example, a circular duct uses 12% less steel than a rectangular duct of the same cross section, provided the velocity of the gas is the same in both cases. In addition, less stiffening is required and wall thickness can be reduced, thereby contributing to lighter support structures and foundations.

Casings are manufactured from carbon steel and internally lined with ceramic fiber insulation.

The insulation is protected by a layer of perforated stainless steel attached to the casing by stainless steel fittings. The lining contributes to sound attenuation.

Interest in CiBAS stems from its simplicity: steam generator, exhaust-gas control damper, and silencer in a single module. It is delivered complete, tested and ready for installation in a day. Size and weight permit mounting directly above a GT. Product’s aerodynamic design minimizes exhaust-gas pressure drop and maximizes power output. Normandale says that the costs of CiBAS units are comparable to those for traditional drum-type boilers because the product’s integral bypass damper permits use of conventional materials for pressure parts rather than expensive alloys.

Two specific turbine types were targeted by this development effort: Rolls Royce plc’s (Mt Vernon, Ohio) RB-211-6761 and Demag Delaval Industrial Turbomachinery Inc’s (Trenton, NJ) GTX 100. The CiBAS products are designed to transfer exhaust heat to a process fluid or to produce steam—superheated or saturated. Up to about 185,000 lb/hr can be produced from the GTX 100 with supplementary firing.

Alstom’s motivations for product development were design and operating limitations of the company’s drum-type boilers for floating production, storage, and offloading vessels (so-called FPSOs)—primarily drum-level control in high seas and weight/footprint limitations offshore. Two units are now operating in the Norwegian sector of the North Sea oil and gas fields where heat captured is transferred to a glycol/water solution. Early this year, the first of three boilers for a refinery in Indonesia started up on the back end of a Hitachi H-25.

Operating experience

When evaluating alternative designs for HRSGs, the due diligence process should include discussions with people who have direct experience operating and maintaining the equipment you are considering. Perhaps the best place to start is to attend an appropriate industry meeting (such as the HRSG User’s Group, and network among your peers. Then follow up with a plant visit.

That’s what the editors of the COMBINED CYCLE Journal did to get a real-world view of how O&M practices for drum-type HRSGs and OTSGs compare. The visit was to Las Vegas Cogeneration LP, operated by Black Hills Generation Inc, Denver. This facility consists of two “plants.” LVC I, more than 10 years old, is an LM6000-powered combined-cycle plant in cogeneration service with thermal energy supplied to a nearby greenhouse; LVC II, in service about three years, consists of two 2 × l LM6000-powered combined-cycle blocks that supply electricity only.

Las Vegas Cogen is one of only a few facilities operating both small drum-type and once-through heat-recovery boilers. Plant I has a conventional dual-pressure Foster Wheeler HRSG that produces about 85,000 lb/hr of 650 psig/750 high-pressure (h-p) steam and approximately 23,000 lb/hr of 105 psig/410F intermediate-pressure (i-p) steam (Fig 9). Plant II has four IST OTSGs, each producing about 100,000 lb/hr of 750 psig/765F and 25,000 lb/hr of 100 psig/420F steam (Fig 10).

Also of interest is that Plant I experiences about 500 starts annually, operating under contract in winter from 5 to 10 a.m. and again from 4 p.m. to midnight; in summer from 10 a.m. to 10 p.m. Its gas turbine is not equipped for power augmentation; fogging and an absorption chiller are able to maintain rated output until ambient temperature exceeds about 95F. The fogging system and chiller may be operated in tandem.

LVC II operates under a tolling contract and typically has been run as a base-load plant for the last year. Its LM6000s are equipped with GE’s EFS (enhanced flow and speed) upgrade to improve low-load efficiency, and with the company’s Sprint power augmentation package, which is used when the turbine is producing more than 90% of its rated output. The EFS enhancement requires more demineralized water—EFS at 18 gpm, traditional Sprint at about 8 gpm. Fogging and mechanical chillers are provided at the GT inlet and, like LVC I, may be operated in tandem.

None of the LM6000s at LVC I or II is equipped with dry low-NOx combustors; none of the heatrecovery steam generators is equipped with duct burners.

David Robb, general manager of the entire Las Vegas Cogen project, O&M Manager Curt Minges, and Senior Operator Richard Winters related how they operate and maintain the two types of boilers and shared some thoughts on what others might do at the design stage to facilitate plant operations. The trio began by saying they have two kinds of starts: hot and cold. Hot refers to any start within 24 hours of shutdown; cold, more than 24 hours.

Almost every startup at Plant I is hot. A typical start goes like this: Push the GT start button. In 15-20 minutes 400-psig superheated steam is available for steam-turbine (ST) warmup. The 10.5-MW ST is rolled at 1000 rpm for around 20 minutes, then brought to full speed and synchronized with the grid. Total time for a hot start is about 45 minutes. On shutdown, vacuum is broken and the ST placed on turning gear. Any oxygen that enters the system is removed by the deaerator at the next start. A cold start takes about an hour and a half.

A startup at Plant II differs because of the OTSGs and the requirement that they be dry during shutdown. Winters explains that the GT exhaust and steam superheat temperatures are the primary variables used to control feedwater flow. H-p feedwater flow is initiated first, but not until the exhaust is at 500F (with minutes of GT fire) and the stack is at 300F. If a startup is within six hours of the previous shutdown, the stack will be at the desired temperature with a minute or two of GT firing. After a 12-hr shutdown, it takes about 15 minutes to reach 300F at the stack; after a shutdown of 24 or more hours, from 20 to 30 minutes.

The h-p circuit must be in full service before water is admitted to the l-p circuit. Ramping h-p feedwater flow from 0 to 100,000 lb/hr takes about 45 minutes. Desuperheating is required during startup because the steam initially is at GT exhaust temperature. Startup of the l-p circuit is achieved within 15 minutes. Note that the h-p circuit is supplied by the boiler-feed pump, the l-p circuit by the condensate pump. At full load, GT exhaust temperature is about 830F; h-p steam is approximately 775F before the attemperator, 760F downstream.

About 20 to 25 minutes after water is admitted to the h-p circuit, steam conditions (about 300 psig/500F) and flow are sufficient to support ST startup. As at Plant I, it takes about 20 minutes after a 12-hr shutdown to complete this procedure and synchronize the unit. The Plant II STs will produce from 20 to 22 MW each on h-p steam only. Rated output (28 MW) is achieved when l-p steam is admitted downstream of the h-p inlet.

Shutdown of a combined cycle equipped with OTSGs is handled this way at LVC II: Condensate flow to the l-p circuit is stopped and feedwater flow to the h-p circuit is reduced to 0 gpm over a period of about 30 minutes. A control algorithm supplied by IST monitors steam temperature in guiding the shutdown process. The ST is shut down at 500 kW, which is about the same time the h-p feedwater control valve is about to close (loss of positive steam production).

The GT remains firing at full speed/no load after the ST is shut down to boil out all the water remaining in the OTSG tubes. (IST says the OTSG does not have to be boiled dry when it is removed from service, unless freezing is an issue or the unit is being prepared for long-term shutdown; simple draining is sufficient.) Boil-out at LVC II takes from 15 to 20 minutes. Steam produced during the boil-out process should flow to the condenser to conserve the high-quality water. However, this condensate rapidly fills the hotwell and unless a storage tank is provided for the excess water it would flood the tube bank region of the condenser. LVC II has no place to put excess condensate so some of the residual steam must be vented to atmosphere.

Once the OTSG is dry, its GT can be shut down and cooled down. Thus shutdown of LVC II takes 45 minutes, or about a half hour longer to accomplish than shutdown of LVC I when both plants are operating in a cycling mode and starting daily.

Minges’ operators say startups and shutdowns of combined cycles with drum-type and once-through boilers are both relatively easy. The longer shutdown associated with the OTSG, however, makes that system more expensive to remove from service than LVC I because of the additional fuel burned during the boil-out process.

Water treatment for LVC II is straightforward despite the high purity requirement for OTSG feedwater. City water is the source of system makeup. The combination of quality raw water and processing by reverse osmosis produces makeup with a conductivity of less than 1 ?S/cm, silica less than 20 ppb, and a pH between 6 and 8. Makeup is stored and, when needed, injected directly into the condenser hotwell. Redundant full-flow, mixed-bed polishers downstream of the hotwell produce high-quality feedwater.

Lessons learned. Robb and his team caution about catalyst selection for systems with OTSGs. A standard catalyst can suffer premature degradation if the gas turbine is operated for too long when the OTSG water circuits are shut down, because of the higher SCR inlet temperature. This is a particularly important consideration with today’s demanding emissions control limits. LVC II, for example, operates under 2 ppm on both NOx and CO. What this means at Robb’s facility is that the LM6000s are restricted to a maximum load of about 20 MW in the simple-cycle mode to ensure an optimum exhaust temperature of less than 780F for proper catalyst operation.

They recommend selecting an SCR catalyst capable of operating safely with inlet temperatures above GT exhaust temperature, Winters while still being most efficient in the desired temperature band. This ensures that simple-cycle operation loads will not be limited strictly because of SCR temperature limitations.

Condensate/feedwater piping selection is another area of concern. At LVC II, all condensate and feedwater piping is carbon steel. To protect this piping, and ultimately the OTSG, from corrosion and the resultant crud, amines and an oxygen scavenger are injected downstream of the mixed-bed polishers. The unwanted effect of this action is an increase in conductivity. Recall that conductivity of OTSG feedwater is limited to 0.25 ?S/cm. Shutdown is required if this level is exceeded.

Another consideration: The cost of demineralization goes up with increasing conductivity. Plus, don’t forget the health and safety concerns associated with the handling of the additional chemicals required for regeneration and the disposal of still more regeneration wastes. A solution: Specify stainless steel for all condensate and feedwater piping in systems with OTSGs. You’ll sleep better.

Emissions monitoring. Designers specified a single-point stack probe for emissions monitoring. But this didn’t work well because of stratification across the large rectangular cross section of the exhaust plenum. The probe supplied was replaced with a multipoint grid. This appears to be giving a much more accurate indication of emissions, even though the stratification issue persists.

Overall performance of the OTSGs has been excellent, according to Robb and his staff. In fact, Plant II availability is in the 98-99% range. However, there was one problem—a tube leak at a failed weld. Normally, this would not be much cause for concern. Leaks on drum-type HRSGs generally are easy to spot on a walk-around inspection because there’s water in most tubes.

But the OTSG is shut down dry and leaks are not easy to find, particularly when you’re not looking for them. But even welds made with a highly automated TIG process and later checked for integrity by a fast UT (ultrasonic) inspection technique can leak on occasion, as Minges’ operators were surprised to learn.

The problem at LVC II was exacerbated by spray from the leak at a U-bend that sensitized adjacent tubes. In effect, the steam/water spray caused a rapid thermal cycling of the material over a relatively long period, weakening the grain boundary. To restore the unit to its original condition, a total of 41 U-bends were replaced. Keep in mind that the TIG welding of Incoloy tubes requires special skills.

IST suggests that the existence of a leak can be identified by an increase in makeup flow or by measuring the differential between feedwater and steam flows. Because OTSGs have no blowdown, makeup should be minimal. What the Las Vegas operators do when the generating units are down for maintenance or other reasons is to pressurize the h-p and l-p OTSG circuits with air and watch for the abnormally fast pressure decay indicative of a leak. CCJ