Careful control of water and steam chemistry at CPI facilities can pay off handsomely in reliable operation
Sometimes lost in the hustle and bustle of production issues at chemical-process-industries (CPI) facilities is that high-purity water, and steam produced thereby, is the lifeblood of the plant. Yet, equipment failures and curtailed production due to water and steam issues can cost a company hundreds of thousands of dollars, or more, on an annual basis. This article outlines methods to improve water and steam operation at chemical and industrial plants.
The following examples come from direct experience on projects for the CPI or related facilities. They outline straightforward issues that may arise at many plants. Following these examples is a discussion of problems and solutions to water and steam chemistry issues.
Example 1. A colleague and I visited an organic chemicals plant in the Midwest, in which the steam superheater bundles in four, 550-psig package boilers had to be replaced every two years or so due to internal scaling. Upon arriving, we were first shown a bundle that had been recently removed, and indeed the internal tube surfaces were covered with a 1/8-1/4 in. scale coating. We then walked around the boilers and immediately noticed foam issuing from the saturated-steam sample lines. Subsequent investigation revealed that total-organic-carbon (TOC) levels in the condensate-return lines to the boilers had been known to reach 200 parts-per-million (ppm). Given that ASME guidelines [ 1] call for a maximum TOC concentration of 0.5 ppm in boilers of this pressure, it was easy to see why so much foam existed in the boiler water, and why impurities carried over to the superheaters on a continual basis.
Example 2. A petrochemical plant in southern U.S. experienced short runs and poor performance of a makeup demineralizer. In particular, the anion resin of the unit underperformed. Analysis revealed that the raw water, which came from an area where rice was grown, contained high concentrations of natural organics. Pretreatment ahead of the demineralizer consisted of clarification and media filtration. Clarification satisfactorily removed suspended solids, but many organics passed through and fouled the anion resin.
Example 3. I have personally witnessed and have heard many reports of plant personnel keeping steam generators online during longterm steam surface condenser leaks. Such leaks, which introduce raw cooling water directly to condensate, cause severe problems in boilers. In one case, in the early 1980s, operation of an 80-MW, 1,250-psig utility boiler for three weeks with a condenser leak resulted in severe hydrogen damage of the boiler tubes. Within a month, tubes began to fail at such a frequent rate that the unit had to be shutdown and entirely re-tubed at a cost of $2 million [ 2].
Example 4. For decades, the conventional treatment program for condensate and feedwater systems called for chemical dosages at sufficient concentration to remove all dissolved oxygen. This regime is now known to be very troublesome and has caused piping and tube failures that have killed at least 10 people in the utility industry in the last decade or so. Industrial boilers, including heat-recovery steam generators (HRSG), are not immune from this attack.
Impurities and their Control
As the introductory examples illustrate, multiple contaminants can enter steam-generation networks from many locations. Impurities include inorganic ions from raw water, such as calcium, magnesium, sodium, chloride, sulfate and bicarbonate, to name the most important, while process operations may introduce these materials and organic compounds to condensate return. Upfront treatment systems to remove these impurities offer the first, and usually most important, line of defense. Makeup-water treatment may range from relatively simple systems for low-pressure steam generation to perhaps reverse-osmosis (RO) and ion-exchange (IX) polishing for high-pressure applications. If process-induced organic compounds are of concern, treatment might include activated carbon filtration. Pre-treatment to protect RO and IX units is often vital, with suspended solids removal by clarification or via the emerging technologies of micro- and ultrafiltration [ 3].
Let’s now look at the condensate feed to the boiler. As a preface to the following discussion it must be noted that when steam generators are first placed in service, steel develops a protective coating of iron oxide known as magnetite (Fe 3 O 4). This mechanism also occurs on boiler waterwall tubes.
3Fe + 4H 2 O → Fe 3 O 4 + 4H 2 ↑ ( 1)
Similarly, copper alloys develop a protective layer of cuprous oxide (Cu 2 O). Formation of protective oxide layers is a very important concept.
Of general importance in condensate and feedwater systems is pH. In piping and heat-exchanger systems of all-ferrous metallurgy, the optimum pH range is 9.2 to 9.6. For networks containing heat exchangers with copper-alloy tubes, the recommended range is a bit lower at 9.1 to 9.3. Control is usually achieved by feed of ammonia or an organic amine to the condensate.
Condensate return and poorly operated steam condensers can introduce large quantities of dissolved oxygen into feedwater systems. The oxygen in turn can cause big problems, as shown in Figure 1.
Methods to control oxygen corrosion primarily comprise the following:
- Mechanical removal of gases in the condenser if the plant has a power-generating steam turbine
- Mechanical removal of gases by a deaerator located in the feedwater system
- Injection of an oxygen scavenger or metal passivator into the feedwater
Mechanical methods for O 2 control. For industrial steam generators that include a steam turbine, the condenser is normally equipped with one or more air-removal compartments for gas removal. Each compartment consists of a shroud, open at the bottom, to which a mechanical vacuum is applied. The vacuum removes air from the condensate, and discharges it to the outside atmosphere. Condenser performance should be monitored on a regular basis, as a crack in the condenser shell or other failure will allow excess air to enter. Air will coat condenser tubes and reduce heat transfer, plus oxygen may enter the condensate.
Downstream from the condenser, the deaerator (DA) provides additional mechanical removal of dissolved gases. As condensate is sprayed or flows through trays downward into the DA-scrubbing vessel, the rising steam heats the condensate and liberates gases. A properly operating DA should reduce dissolved oxygen concentrations down to 7 ppb or less.
Chemical control of O 2. Into the 1980s, in the U.S. at least, conventional wisdom called for complete removal of oxygen from feedwater. It was thought, incorrectly as we shall see, that the total absence of O 2 was best for corrosion control in the feedwater network. Thus, mechanical O 2 removal was supplemented with chemical treatment. One of the first practical chemicals to be employed in steam generators was sodium sulfite (Na 2 SO 3). Sodium sulfite reacts with oxygen to produce sodium sulfate:
2Na 2 SO 3 + O 2 → 2Na 2 SO 4 ( 2)
The primary advantages of sodium sulfite are that it is a common and inexpensive chemical, is non-toxic, and can be used to treat boiler water in which the steam is extracted for food processing or other applications regulated by U.S. Food and Drug Administration (FDA). Sodium sulfite is primarily used in low-pressure industrial boilers (<600 psig), because it adds too many dissolved solids to high-pressure boiler water. Also, in boilers that operate above 900 psig, sodium sulfite will thermally decompose to produce hydrogen sulfide (H 2 S) and sulfur dioxide (SO 2), both of which are quite corrosive.
For industrial boilers that operate at pressures above 900 psig, alternative chemicals are more suitable for oxygen scavenging. The workhorse for many years was hydrazine (N 2 H 4), which reacts with oxygen as follows:
N 2 H 4 + O 2 → 2H 2 O + N 2 ↑ ( 3)
Hydrazine is advantageous because it does not add any dissolved solids to the feedwater, it reacts with oxygen in a one-to-one weight ratio, and it is supplied in liquid form at 35% concentration.
A primary benefit of hydrazine is that it will passivate oxidized areas of piping and tube materials as follows:
N 2 H 4 + 6Fe 2 O 3 → 4Fe 3 O 4 + N 2 ↑ + 2H 2 O ( 4)
Similarly, if the heat exchangers have copper-alloy heat-exchanger tubes, the chemical will return oxidized layers of cupric oxide (CuO) to the more stable cuprous oxide (Cu 2 O).
N 2 H 4 + 4CuO → 2Cu 2 O + N 2 ↑ + 2H 2 O ( 5)
Hydrazine residuals were typically maintained at relatively low levels of perhaps 20-100 ppb.
Given the simplified chemistry of hydrazine, it would appear to be the ideal oxygen scavenger. Unfortunately, hydrazine is considered to be a potential carcinogen and is now registered as a hazardous compound. Handling procedures have become very stringent. This difficulty has, in part, led to the development of other treatment chemicals. Many of the major water-treatment companies provide alternative oxygen scavengers, some principal ones being hydroquinone, carbohydrazide (N 4 H 6 CO), and methyl ethyl ketoxime (C 4 H 9 NOH). All of these products also passivate metals.
Safe hydrazine systems. It is possible to set up a hydrazine feed system that does not expose workers to the compound. In one such system at a western-U.S. utility, the hydrazine solution is supplied in portable, reusable containers. Personnel connect each new container to a permanent metering pump and distribution line, which transport precise dosages of the chemical in neat form to the feedwater system. A variation of this concept is a system in which the primary feed tank supplies a closed day tank, which is vented to the outside atmosphere. The operator introduces a measured volume of hydrazine to the day tank, followed by dilution water. A metering pump feeds the solution into the system.
The use of a pH conditioner and oxygen scavenger (metal passivator), particularly the latter, constitutes what is known as an all-volatile reducing AVT(R) program in condensate and feedwater systems. In the 1980s and 1990s, researchers began to discover that AVT(R) was the cause of previously unspecified problems. Most notable is the dissolution of magnetite at a temperature range and chemical conditions common to the feedwater network. This behavior is shown in Figure 2.
As the title of Figure 2 suggests, pH in high-purity condensate and feedwater is typically a function of the ammonia concentration. It is the lower pH, at low ammonia concentrations in a reducing environment, which is responsible for magnetite dissolution. This explains why corrosion can be much higher at an NH 3 concentration of 0.1 ppm than in any other case. The ammonia does not attack the magnetite directly. So, in a reducing environment established by feed of an oxygen scavenger (metal passivator), gradual magnetite dissolution has led to sudden, and in some cases fatal, failures of mild steel by flow-accelerated-corrosion (FAC). FAC develops at flow disturbances and direction changes, for example at feedwater and economizer elbows, reducers and tees, in strongly reducing environments. The environment causes continual leaching of ferrous ions (Fe +2) from the tube surface that weakens the wall structure and eventually reduces pipe strength to the point of sudden failure. An example of FAC is shown in Figure 3.
FAC has been a problem in heat-recovery steam generator (HRSG) waterwall tubes, which have many tight-radius elbows. The low-pressure circuits of HRSGs often operate near the temperature of highest corrosion potential (shown in Figure 2), which further exacerbates the issue. A particular difficulty with HRSGs is that the two or three semi-independent waterwall circuits make chemistry control rather difficult. One solution, albeit at some cost, for controlling FAC in HRSGs in the design phase is to specify tube material, at least in elbows, of 1.25% chrome steel. This material is resistant to attack.
The next sections outline chemical techniques for combating FAC, while preventing other corrosion in feedwater systems.
Oxygenated treatment (OT). OT is a feedwater treatment developed over 30 years ago for the utility industry. In an OT program, oxygen is deliberately introduced to the condensate and feedwater system. Two variations of oxygenated treatment are most popular. In the first, O 2 is injected alone without any pH-conditioning chemicals. This program is known as neutral water treatment (NWT). More often, ammonia is also injected for pH control. This is known as combined water treatment (CWT).
Both NWT and CWT require the controlled injection of O 2 into the condensate and feedwater system. Properly applied, OT establishes a surface layer in part composed of ferric oxide hydrate (FeOOH), which is more stable than magnetite. Typical injection points are just after the condensate polisher and again at the deaerator outlet (Figure 4). In CWT programs, which are most common in the U.S., oxygen is dosed to maintain a 30 to 150 ppb residual. Ammonia is added to raise the pH within a range of 8.0 to 8.5. Typically, 20 to 70 ppb of ammonia will control water chemistry within this pH range.
The keys to an OT program are controlled oxygen feed and high-purity condensate, where cation conductivity can be maintained at ≤0.15 microSiemens/cm (µS/cm). OT is probably too elaborate for low-pressure industrial boilers, but it has been applied to some utility drum units and to HRSGs. OT cannot be used in systems that contain copper-alloy feedwater heater tubes, as copper corrosion would be much too severe.
AVT(O). An offshoot of OT is a program developed by the Electric Power Research Institute (EPRI) known as all-volatile treatment (oxidizing), or its acronym AVT(O). The idea continues to be the establishment of a FeOOH layer on the feedwater piping, but by a less intensive mechanism. What the researchers found is that, in condensate and feedwater networks where air seepage into the condenser (air in-leakage) is minor and condensate dissolved oxygen levels stay at or below 10 ppb, discontinued feed of an oxygen scavenger allows the FeOOH protective layer to form naturally. As with OT, this program is only applicable in systems with all-ferrous metallurgy. One major difference from OT is that the pH should be maintained within a range of 9.2 to 9.6. An operating guideline is cation conductivity ≤0.2 µS/cm. Excursions in dissolved-oxygen concentration and cation conductivity, particularly the former, indicate excess air in-leakage within the condenser. Increased air in-leakage also introduces excess carbon dioxide, which influences corrosion. Thus, for a unit on AVT(O), any air leaks into the system that raise condensate dissolved-oxygen levels much above a mid-teen ppb concentration should be investigated and corrected as quickly as possible.
AVT(R) for mixed-metallurgy systems. For mixed-metallurgy systems, OT and AVT(O) are not acceptable, as they would initiate and sustain excessive copper corrosion. However, operation with complete removal of oxygen leads to FAC. So, the correct program is to feed an O 2 scavenger, but at reduced concentrations to minimize FAC. Chemistry control can be quite difficult when relying upon standard dissolved-oxygen and oxygen-scavenger analyses. The technique of oxidation-reduction-potential (ORP) monitoring is becoming popular for mixed-metallurgy condensate/feedwater chemistry. In short, online ORP monitors measure the electrochemical potential of the solution versus a standard electrode, most commonly Ag/AgCl, saturated KCl. A general rule-of-thumb is that the O 2 scavenger should be fed to maintain an ORP within a range of – 350 to – 300 mV. This corresponds to a range of – 150 to – 100 mV for a standard hydrogen electrode (SHE). However, chemists have found that this guideline should not be considered an absolute [ 4]. A better plan is to set up comprehensive tests that include dissolved iron and copper analyses, and coordinate the optimum ORP range with minimized copper and iron concentrations.
Raw makeup water or heat-exchanger-cooling water from a lake or river typically contains several hundred ppm of cations and anions, most notably calcium, sodium, magnesium, potassium, bicarbonate, chloride, silica, and sulfate, as well as other materials including suspended solids. As these contaminants enter the boiler, a number of temperature-induced reactions will occur. Two common reactions are shown below.
Ca +2 + 2HCO 3 -1 → CaCO 3 ↓ + CO 2 ↑ + H 2 O ( 6)
Ca +2 (or Mg +2) + SiO 3 -2 → CaSiO 3 ↓ (or MgSiO 3 ↓) ( 7)
Equations (6) and (7) are typical scale-forming reactions. Even a relatively thin deposit layer will significantly reduce heat transfer, and a boiler must be fired harder to achieve the same level of steam production. This in turn can lead to overheating of the boiler tubes, which will shorten tube life.
Much more problematic, particularly in high-pressure boilers of at least 1,000 psig or higher, is the effect that cooling-water leakages into the system have with regard to rapid and potentially catastrophic corrosion. The reaction shown below is a prime example.
MgCl 2 + 2H 2 O → Mg(OH) 2 ↓ + 2HCl ( 8)
As can clearly be seen, a product of this reaction is hydrochloric acid. While HCl may cause general corrosion in and of itself, when concentrated under deposits, the acid reacts with iron to generate 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 and react with carbon atoms in the steel to generate methane:
2H 2 + Fe 3 C → 3Fe + CH 4 ↑ ( 9)
Formation of the gaseous methane and hydrogen molecules causes cracking in the steel, greatly weakening its strength. Hydrogen damage (Figure 5) 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.
Condensate return may carry any number of contaminants, including organic chemicals as mentioned in the first example. Organics can bake onto tube surfaces, and they will also cause foaming in boiler drums, which in turn leads to carryover of impurities to steam. The superheater failures mentioned in Example 1 are one possibility, but if the system produces power with a steam turbine, the impurities can settle out on turbine blades and cause additional corrosion.
Modern technologies offer techniques to combat organic fouling in condensate return. Space limitations prevent an in-depth discussion of these technologies in this article, but Reference [ 5] outlines a modern system to control organics. The system utilizes adsorption cartridges and bioreactors to remove organics, even from refinery condensate-return streams, such that the purified water reportedly contains low-ppb concentrations of carbon-based compounds. Successful operation of this type of equipment would certainly have reduced the headaches encountered by the personnel in Example 1.
Treatment by Phosphate
For over seven decades, steam-generation chemists have utilized sodium phosphate compounds for corrosion control and prevention of solids deposition in the waterwall circuits of drum-type, steam-generating systems. Alternatives are also available, but can be tricky to control.
In the early days of steam generation for power production, various odd-sounding treatment methods were used. Some of these treatments had operators putting sawdust or potato peels into boilers. These natural products contributed large organic molecules such as lignins, tannins, and starch to the water. The chemicals sequestered hardness ions. Other chemistry programs actually allowed the formation of calcium carbonate on tube walls to protect the tube surface from the boiler water.
In the 1930s, researchers developed the technique of adding tri-sodium phosphate (Na 3 PO 4) to boilers to generate alkaline conditions.
Na 3 PO 4 + H 2 O → Na 2 HPO 4 + NaOH ( 10)
In early, low-pressure boilers, phosphate concentrations in a range of 20-40 ppm were common. However, as higher-pressure boilers came on line, tubes began to fail due to the phenomenon of caustic gouging (Figure 6).
Research showed that this failure occurred underneath boiler tube deposits. Figure 7 illustrates a porous tube deposit where water penetrates the deposit through various channels. As the water approaches the tube surface, temperatures increase. The water boils off, leaving other species behind. This phenomenon is known as wick boiling. In the case of early phosphate programs, and to a lesser but not unknown frequency today, sodium hydroxide remains after wick boiling. Concentrations may rise to levels many times that in the bulk boiler water. The concentrated NaOH attacks the boiler metal and protective magnetite film via the following reactions:
Fe + 2NaOH → Na 2 FeO 2 + H 2 ↑ ( 11)
Fe 3 O 4 + 4NaOH → 2Na 2 FeO 2 + N 2 FeO 2 + 2H 2 O ( 12)
Because phosphate serves another valuable function besides pH control, programs were not discontinued but rather modified to current regimes where free NaOH alkalinity is limited to 1 ppm, with phosphate concentrations of 0.2 to 3 ppm in the low range and 3 to 10 ppm for those who feel comfortable with more protection.
Control of contaminant in-leakage, at least until repairs can be made, constitutes a phosphate program’s second major function. Phosphate reacts directly with calcium to produce calcium hydroxyapatite:
10Ca +2 + 6PO 4 -3 + 2OH -1 → 3Ca 3 (PO 4) 2 .Ca(OH) 2 ↓ ( 13)
Magnesium and silica react with the alkalinity produced by phosphate to form the non-adherent material, serpentine:
3Mg +2 + 2SiO 3 -2 + 2OH -1 + H 2 O → 2MgSiO 3 .Mg(OH) 2 .H 2 O↓ ( 14)
Calcium hydroxyapatite and serpentine exist as soft sludges and are much easier to remove than the hard scale or corrosive products that would otherwise form. They typically settle in the mud drum or lower headers from which they are removed by blowdown.
Edited by Gerald Ondrey