Emissions

Critical HRSG

Combined-cycle power plants continue to be the primary replacement for the many coal-fired plants that have been or soon will be retired. 

On-Line Water/Steam Monitoring Parameters

By Brad Buecker

Combined-cycle power plants continue to be the primary replacement for the many coal-fired plants that have been or soon will be retired. Like their coal boiler predecessors, combined-cycle heat recovery steam generators (HRSGs) operate at high pressures and temperatures. Proper chemistry control and monitoring are critical to prevent corrosion and fouling upsets that may range from relatively minor to catastrophic.

Schematic of a triple-pressure, FFLP HRSG 1

Two examples of the latter, where some past failures have caused fatalities, are release of high-temperature, high-pressure feedwater from flow accelerated corrosion (FAC), and turbine blade failure (with the turbine spinning at 3,600 rpm) due to previous corrosion mechanisms. Continuous, on-line monitoring of complete circuit chemistry is essential for reliable and efficient operation. Yet, in this author’s experience of reviewing specifications for many of these projects, it has become obvious that plant designers often have difficulty separating important measurements from others that are of little value. This article provides an overview of the critical analyses that should be part of any HRSG water/steam chemistry monitoring system. Much more detailed information is available from the organizations listed later in this document.

Following Chemistry Through the HRSG Circuit

A number of different HRSG designs are possible, and space limitations prevent a chemistry discussion of these variations. For this article, we will utilize a very common design shown schematically below.

This design is termed, as can be found in the literature [1], a “feed forward low-pressure (FFLP)” HRSG, where all of the feedwater passes through the low-pressure HRSG evaporator for pre-heating. A small amount of steam is taken off the LP drum, but the vast bulk of the fluid is distributed to the high-pressure (HP) and, to a much lesser extent, intermediate-pressure (IP), evaporators. Also note that this schematic includes an external deaerator (DA). In many designs, the DA is integral to the LP drum. But, the issue of deaeration brings up a more important point, where advancements in chemistry control have in many cases eliminated the need for a deaerator. The following discussion does not include deaeration.

Chemistry control and monitoring have four primary goals during normal operation:

  • Protect the entire steam generation network from general corrosion.
  • Protect those areas (feedwater system, economizer, low-pressure evaporator) that are most susceptible to flow-accelerated corrosion.
  • Minimize impurity ingress (the condenser is the prime location), which can cause severe localized corrosion, particularly in the HRSG evaporators.
  • Protect the steam turbine from transport of impurities that can induce corrosion and deposition.

Also very important but often overlooked is corrosion protection when the unit is off-line. The author will hopefully be able to re-address this subject in a future article for Power Engineering.

We will begin our review of important monitoring details and chemistry control with the makeup water treatment system and then progress forward.

Makeup Water System

High-purity water is a requirement for high-pressure steam generators. Very common in the industry now is a core treatment method of two-pass reverse osmosis (RO), with polishing of the RO effluent either by portable mixed-bed (MB) ion exchangers or electrodeionization (EDI). Excluding RO and EDI instrumentation from this discussion, the recommended continuous readings and normal limits of the system effluent are:

  • Specific conductivity: ≤0.1 μS/cm
  • Sodium: ≤2 parts-per-billion (ppb)
  • Silica: ≤10 ppb

Specific conductivity provides a blanket measurement of all impurities in the stream. Sodium is the most weakly held ion on the cation exchange resin of a mixed-bed unit, and is the first cationic constituent to come off if the resin reaches exhaustion. Silica is the most weakly held constituent on the anion resin, thus anion effluent silica analyses are very helpful for monitoring bed performance and exhaustion.

The author has seen many specifications that also call for pH measurement of makeup system effluent. This analysis is essentially useless for two reasons. First, in high-purity water without chemical treatment, pH is quite difficult to measure. Second, the concentration of either hydrogen ions (H+) or hydroxyl ions (OH-), which give water either acid or basic properties, is so small that corrosion from either of these constituents is not an issue.

Condensate Storage Effluent

At the plant level, it is still not often well understood that severe corrosion can occur in steam generators during outages. Due to the nature of combined-cycle operation, many facilities start up and shut down numerous times during the course of the year. To these cycles can be added maintenance outages, which obviously may be of much longer duration. If the units are allowed to cool, dissolved oxygen (D.O.) ingress can be a significant problem. One location of high D.O. is the condensate storage tank, which is typically vented to the atmosphere. Any filling of the HRSG, partial or full, from this tank introduces to the unit cool condensate saturated with oxygen. The reader is encouraged to search the Power Engineering archives for reference 2, which outlines the use of gas transfer membrane (GTM) technology for dissolved oxygen removal from condensate along with other methods, nitrogen blanketing and dehumidified air circulation, for layup protection of HRSGs. The GTM system outlined in the article consistently produces condensate with a D.O. concentration of less than 10 ppb. Reference 2 also discusses the layup techniques of nitrogen blanketing and dehumidified air circulation, which, if applied properly, are extremely valuable in minimizing off-line corrosion.

Analysis of a Failure 2

Condensate Pump Discharge

Condensate pump discharge (CPD) is an absolutely vital monitoring point, particularly for HRSGs with steam surface condensers. Condenser tube leaks introduce cooling water, with all of its impurities, to the condensate and of course then to the steam generator. There, the high boiler temperatures will induce reactions that have severe corrosion and scale-forming effects. We will examine some of this chemistry in a bit more detail later.

Recommended on-line analyses include:

  • Cation (or degassed cation) conductivity: ≤0.2 μS/cm
  • Specific conductivity: Consistent with pH generated by ammonia feed
  • pH: For FFLP HRSGs on boiler water phosphate treatment, range of 9.2 to 9.8
  • Sodium: <2 ppb
  • Dissolved oxygen: ≤20 ppb

Evaluation of cation conductivity, specific conductivity, and pH are linked. In all steam generators, the pH is typically maintained in a mildly alkaline range, as this minimizes general corrosion of carbon steel. For FFLP HRSGs, the recommended pH range is 9.6 to 10.0 to protect the feedwater piping and LP economizer and evaporator tubes from both general corrosion and another mechanism, flow accelerated corrosion (FAC). We will examine FAC in a bit more detail in the next section. Ammonia (sometimes an amine) is the common chemical of choice for pH control.

NH3 + H2O ⇔ NH4+ + OH-

Experience has shown that control of ammonia feed is easier if it is based on specific conductivity measurements rather than pH. So, during initial unit startup the correlation between S.C. and pH is established, with subsequent ammonia feed derived from S.C. readings.

So then what is the link to cation conductivity (C.C.)? Specific conductivity is influenced by all ions, including the ammonia introduced for pH control. Thus, S.C. is not effective for detecting impurity ingress, especially if a condenser tube leak or other source of contamination is slight. With cation conductivity, the sample stream passes through a cation exchange column that exchanges ammonium (NH4+) and any other cations, e.g., sodium, calcium, etc. for hydrogen ions (H+). The hydrogen ions combine with anions in the stream, primarily chloride and sulfate, which have remained untouched by the exchange column. Thus, the effluent from the column is a dilute stream of acids, primarily HCl and H2SO4. These acids have a greater conductivity than their salt counterparts, and so C.C. can be an excellent tool for monitoring impurity ingress. However, C.C. is influenced by carbon dioxide, which may reach measureable proportions if air in-leakage at the condenser is significant. Degassed cation conductivity (D.C.C.) employs a reboiler or perhaps a nitrogen purge vessel to “gas off” most of the CO2 prior to the conductivity measurement. Thus, D.C.C. is a more true measurement of anions, except for one factor which on occasion is important. If an amine (an organic ammonia compound) is employed in place of ammonia for pH control, all conductivity readings may become meaningless. With variability due to the amine type, at least some of any of the typical neutralizing amines will carry over from boiler water into the steam. At the very high steam temperatures, the amine will decompose to small-chain organic acids, primarily acetate and formate, and perhaps some CO2. These acids mask the conductivity of other constituents. The decomposition products are not removed by cation exchange columns or degassing methods.

Sodium monitoring is an important supplement to cation conductivity. All cooling water contains sodium, and it is a very easy element to monitor on-line. With a tight condenser, sodium levels in the condensate should be very low (<2 ppb), and in many cases less than 1 ppb. Sodium and cation conductivity measurements in parallel provide a quick indication of impurity ingress. And, this combination of instruments is effective for differentiating between a true upset and simply instrument error. Say for example that cation conductivity suddenly increases above 0.2 μS/cm but sodium remains steady. C.C. instrument error or a needed calibration is the likely culprit.

Measurement and control of dissolved oxygen has evolved greatly in the last two decades or so, which we will explore in the next section. D.O. measurement at the condensate pump discharge is important for evaluation of air leakage into the condenser. As we shall see, some oxygen in the condensate is essential for control of flow-accelerated corrosion.

Feedwater/ Economizer Inlet

This sample is very important because it is the last checkpoint before the steam generator, and is downstream of the chemical injection point. Also, this is the water used for steam attemperation. Many of the feedwater guidelines mirror or are similar to those of the condensate pump discharge, as, unlike the coal-fired units of the past, HRSGs do not have feedwater heaters. For FFLP HRSGs, feedwater values also represent the chemistry of the LP circuit.

Recommended on-line feedwater analyses include:

  • Cation (or degassed cation) conductivity: ≤0.2 μS/cm
  • Specific conductivity: Consistent with pH generated by ammonia feed
  • pH: For FFLP HRSGs, range of 9.6 to 10.0
  • Sodium: <2 ppb
  • Dissolved oxygen: 5 to 10 ppb
  • Iron: <2 ppb

The focus in this section is on D.O. and iron analyses.

When steam generators are first placed into service, the carbon steel feedwater piping and steam generator waterwall tubes develop a surface layer of magnetite (Fe3O4). Likewise, for any systems with copper alloys, the alloy forms a protective cuprous oxide (Cu2O) layer. With respect to steel, if uncontrolled oxygen enters the system, it can oxidize the magnetite to rust (Fe2O3), with severe corrosion consequences. For many years boilers were designed, and plant chemists were trained, to remove all traces of dissolved oxygen from boiler feedwater. Mechanical deaerators could typically reduce the D.O. concentration to 7 ppb, with the remainder being eliminated by feed of a reducing agent, aka oxygen scavenger, such as hydrazine. The combination of ammonia feed for pH control and reducing agent feed for D.O. control became known as all-volatile treatment reducing [AVT(R)].

But, it has now become relatively well known (albeit not known nearly well enough) that these conditions can generate the phenomenon of single-phase flow-accelerated corrosion (FAC) of carbon steel. FAC occurs at flow disturbances, e.g,. economizer and evaporator elbows, control valves, etc. The combination of the chemical environment and fluid impingement induces iron dissolution from the impacted location. Gradual wall thinning is the result.

The corrosion rate is also influenced by temperature (maximum near 300oF) and pH (corrosion increases with decreasing pH, particularly below 9.0). What chemists have discovered, and whose impetus came from research in the 1970s in Europe and Russia on supercritical boilers, is that with very pure water, a small but continuous pure oxygen feed to the condensate and feedwater of the units causes the magnetite layer to become interspersed and covered with the iron oxide FeOOH, which forms a dense and strong layer. This program is known as oxygenated treatment (OT). Scientists from EPRI (Electric Power Research Institute) modified this program for drum units, where the oxygen is introduced to the condensate from natural condenser air in-leakage. This treatment, combined with ammonia feed but no oxygen scavenger, will establish similar FeOOH surface chemistry on carbon steel. The program is known as all-volatile treatment oxidizing [AVT(O)]. A requirement is condensate/feedwater cation conductivity ≤0.2 μS/cm. The dissolved oxygen requirement explains the 20 ppb D.O. limit in the condensate pump discharge and the 5 to 10 ppb range in the feedwater outlined above. If the feedwater range cannot be established, it may be necessary to feed a small amount of pure oxygen to the feedwater. For the FFLP shown in Figure 1, the injection point would be at the boiler feed pump suction, downstream of the LP evaporator. AVT(O) requirements also explain the decreased importance of mechanical deaeration for HRSGs.

Iron monitoring is highly recommended to verify that the treatment program is controlling FAC. Improved instrumentation, primarily centered upon real-time particulate monitoring and corrosion product sampling, is available to track condensate/feedwater iron concentrations. An extremely important issue regarding corrosion of condensate/feedwater piping is that the iron oxide corrosion products travel to the steam generator and precipitate primarily on the hot side of the waterwall tubes. These porous deposits can induce chemical reactions underneath the deposit that might not occur at all in the bulk boiler water.

The reader will note no reference to copper in the above discussion. Very few if any HRSGs have copper alloys anywhere in the condensate/feedwater system (copper-alloy condenser tubes do not count). If, for some reason, copper alloys are present, AVT(O) cannot be used. Rather, AVT(R) must be employed with very careful monitoring of iron and copper corrosion.

Even though FAC is well known in some circles, the author regularly sees HRSG specifications that include language requiring oxygen scavenger feed. It often takes protracted conversations with developers and owner’s engineers to explain that this chemistry is no longer valid for condensate/feedwater systems that contain no copper alloys. Lest any readers not believe this recommendation, the author invites them to refer to EPRI or International Association for the Properties of Water and Steam (IAPWS) documents for further confirmation. These documents also outline in detail the mechanism of two-phase FAC, which, as its name implies, may occur in spots where a combination of steam and water exist. In HRSGs, the LP drum is the most likely spot for two-phase FAC. Deaerators and feedwater heater drains in conventional units are other likely locations.

Boiler Water

The reader will note that the analysis list below includes only a few numeric guidelines. This is because in drum units the allowable concentrations of impurities are quite dependent upon boiler pressure. EPRI, IAPWS, and others have developed charts and guidelines that illustrate limits as a function of steam generator pressure. The values in those documents should be used as an initial baseline with subsequent clarification after unit startup. In very large measure, boiler water monitoring is necessary to protect steam purity.

The recommended boiler water analyses do not differentiate between HP and IP evaporator circuits, but it should be recognized that HP is more critical due to the higher pressure and temperature.

  • pH: Immediate unit shutdown if the pH drops below 8.0. Recommended range for phosphate-treated units, 9.2 to 9.8
  • Cation conductivity
  • Specific conductivity
  • Chloride
  • Silica
  • Sodium
  • Phosphate (for those units on phosphate treatment)

Boiler water chemistry is much more complex than many people understand. The most important analysis is pH. But, an accounting is needed for the influence of ammonia that transports into the HRSG HP evaporator from the feedwater if tri-sodium phosphate in particular is utilized for boiler water treatment. Computer programs are available to do this. [3] Such programs incorporate cation conductivity and specific conductivity measurements to accurately calculate impurity concentrations, including the worst of the bad actors, chloride. But now, on-line chloride analyzers are available for direct monitoring. A brief discussion of this chemistry is in order at this point. A large condenser leak or chronic small leaks can introduce chlorides at a level that may overwhelm the neutralizing capability of phosphate or caustic treatments. Chlorides, like other impurities, can concentrate underneath porous iron oxide deposits on the waterwall tubes. The following equation outlines the reaction that then may occur at the tube surface.

MgCl2 + 2H2O + heat → Mg(OH)2↓ + 2HCl

As can clearly be seen, a product of this reaction is hydrochloric acid. While HCl can cause general corrosion in and of itself, the compound will concentrate under deposits, where the reaction of the acid with iron generates hydrogen, which in turn can lead to hydrogen damage of the tubes. In this mechanism, hydrogen gas molecules, which are very small, penetrate into the metal wall where they then react with carbon atoms in the steel to generate methane (CH4), a large molecule:

2H2 + Fe3C → 3Fe + CH4↑

Formation and intrusion of the gaseous methane and hydrogen molecules causes cracking in the steel, greatly weakening its strength.

Hydrogen damage is very troublesome because it cannot be easily detected. After hydrogen damage has occurred, the plant staff may replace tubes only to find that other tubes continue to rupture. [4]

Phosphate chemistry is still common, but only tri-sodium phosphate (Na3PO4) is recommended as the treatment chemical. (The old, discredited coordinated and congruent phosphate programs utilized a blend of TSP and di-sodium phosphate, with perhaps an occasional pinch of mono-sodium phosphate.) TSP is more robust than ammonia in generating alkalinity.

Na3PO4 + H2O ⇔ Na2HPO4 + NaOH

However, even a TSP program can be problematic due to the peculiar solubility characteristics of TSP.

At the high-temperatures of utility steam generators, most of the phosphate precipitates on boiler internals during normal operation. A reduction in unit load or shutdown allows the phosphate to re-dissolve. This “hideout” plays havoc with chemistry control, and particularly for combined-cycle units that cycle regularly. Continuous phosphate analyses rather than grab sample analyses are much better for tracking hideout. Hideout issues have pushed a number of plant chemists to abandon phosphate treatment in favor of caustic treatment or perhaps even AVT. The latter though offers virtually no protection against contaminant ingress. For any units that are operated on AVT, and really for any units at all with steam surface condensers, condensate polishing is recommended to protect the steam generators. Yes, condensate polishing adds to upfront equipment costs and requires extra labor, but the safety a polisher provides can be enormous.

Silica is a compound that will carry over vaporously into steam, which makes monitoring quite important in the boiler water and especially the steam. For example, in a 1000 psi boiler the recommended maximum drum water silica concentration is 2.3 ppm. In a 2,400 psi steam generator the recommended maximum is 0.22 ppm! But again to re-emphasize, the allowable limits for silica and many of the other impurities are based on protection of steam purity. The following analyses are critical in that regard.

Steam

Main and reheat steam samples are the most important of all steam samples because the analyses reflect the total contaminant transport to the turbine, both from carryover in the drums and from direct introduction by attemperator water.

Recommended on-line analyses are:

  • Cation conductivity (or D.C.C.): ≤0.2 μS/cm
  • Sodium: ≤2 ppb
  • Silica: ≤10 ppb
  • Chloride*: ≤2 ppb
  • Sulfate*: ≤2 ppb

Cation conductivity is basically a blanket measurement of steam purity. It does not provide any speciation of impurities. But, C.C. is the criteria that turbine manufacturers often select for equipment commissioning. New evidence suggests a more realistic D.C.C. limit of 0.1 μS/cm to better protect turbines. [6] Sodium is an important monitoring parameter, not only for possible salt carryover but also for the potential of sodium hydroxide transport to the turbine. With a well-operated unit, the sodium concentration should typically be well below 2 ppb, and often significantly below 1 ppb. Silica monitoring is of course necessary to keep track of vaporous carryover of this compound, which can deposit on turbine blades.

The reader will note the asterisk by chloride and sulfate. These are important parameters, as it is the salts of chloride and sulfate (and sodium hydroxide if too much caustic is allowed to accumulate in the boiler water) that precipitate on the last stages of the LP turbine and then cause corrosion during outages. In the past, the only accurate method for trace chloride and sulfate analyses was by ion chromatography (IC). As the author knows from direct experience, IC can provide good data in the hands of a skilled operator, but the technology is expensive and requires significant attention. Very recently a new analytical technique has been introduced, [7] and has been successfully tested. The instrument requires virtually no maintenance and in a test the author helped arrange provided accurate readings for both Cl and SO4 down to 0.2 ppb. The technology shows excellent promise for trace monitoring of these most problematic impurities.

Monitoring guidelines typically also call for saturated steam analyses for C.C., sodium, and silica. These can be valuable for evaluating carryover from the boiler drums, but saturated steam quickly becomes a two-phase fluid in the sampling line, which makes representative sampling somewhat difficult.

One final comment, the reader will note in the reference list mention of the Electric Utility Chemistry Workshop and the International Water Conference. These are both excellent annual events for acquiring technical information and for networking with industry colleagues.

Author

Brad Buecker is a process specialist with Kiewit Power Engineers in Lenexa, Kan., and a contributing editor for Power Engineering.

 

Tags