A Specifying Engineer’s Guide to Water Treatment

Brad Buecker, Process Specialist, Kiewit Engineering and Design Co. | February 09, 2016

Many requests-for-proposals (RFP) for new combined-cycle power plants contain a flaw. In the majority of cases, it is apparent that the project developers (and often the owner’s engineer) do not fully understand the importance of water treatment and steam generation chemistry. Often, insufficient raw water quality data is provided at the outset. This makes precise design of the makeup water system difficult or impossible to achieve. Also, in the competitive business of bid preparation, projects often are awarded to the low bidder without sufficient thought given to whether the design or equipment offering is satisfactory. Then, after commissioning, and when systems underperform or even fail, the owner and operators are placed in a severe bind.

What’s more, many RFPs specify out-of-date or discredited chemical treatment programs. A primary example for heat recovery steam generators (HRSGs) is a continued belief in the need for oxygen scavengers (a more accurate term is reducing agent) as a feedwater treatment. This article explores some of these specifying issues.

Makeup Water Treatment Design

Fresh water supplies are no longer ubiquitous for power plant makeup. Increasingly, plants have to use treated municipal wastewater, poor quality groundwater or other less-than-ideal supplies. Raw water quality greatly impacts the design, and often the sizing or redundancy, of the makeup water treatment system and pre-treatment equipment. Here are a few of the constituents in raw water that must be accounted for in system design:

The hardness ions, calcium and magnesium (Ca and Mg). These can react with alkalinity and silica in the water to form scale in reverse osmosis (RO) systems and ion exchange units.
Bicarbonate alkalinity (HCO₃^-). As the concentration increases in an RO unit or as temperature increases in a heat exchanger, alkalinity can react with calcium to form calcium carbonate (CaCO₃) deposits.
Silica (SiO₂). Silica chemistry is complex. Silica can form scale, but more often combines with magnesium and sometimes calcium to form silicate scales. These are very difficult to remove. Silicate scale formation becomes more pronounced with increasing pH. However, if no hardness ions are present, higher pH will keep SiO₂ in solution. The latter chemistry is the basis for a membrane-based wastewater treatment technology that is becoming more popular. [1]
Chloride (Cl). Chloride is a notorious pitting agent of stainless steels, especially underneath deposits.
Sulfate (SO₄). Sulfate will combine with calcium to form deposits, although the solubility of CaSO₄ is considerably higher than CaCO₃. However, the deposits are difficult to remove. Sulfate will also form tenacious scales, particularly in RO systems, with barium and strontium. These metals typically exist in trace quantities in raw water, but, if present in large enough concentrations, can cause problems.
Iron and manganese (Fe and Mn). These metals exist in a number of valence states. If they enter treatment systems in dissolved form, they can cause fouling and sometimes corrosion.
Suspended solids can be death to RO membranes and must be removed prior to RO treatment.

As a result, raw water analyses for any new project should include these elements or compounds along with pH, total dissolved solids (TDS), phosphorus, fluoride, ammonia, oil and grease, and total organic carbon (TOC). Also, for any system that uses reverse osmosis, silt density index (SDI) tests are a requirement for the RO feed. A single snapshot analysis set is not sufficient; rather, multiple and historical data are needed.

During large power plant construction in the last century, a common treatment design was clarification/filtration, followed by ion exchange. While this technology was often effective, problems sometimes arose. Changes in flow, temperature and other factors caused solids carryover from clarifiers, which, in turn, induced downstream fouling. Clarifiers were large structures, often circular or cone-shaped, that had broad footprints. A common design criteria for clarifiers is the rise rate, which is gallons perminute (gpm) of flow divided by the area of the water surface. A reasonable rise rate for older clarifiers is 1 gpm/ft². Modern systems may now be able to achieve rise rates of 25 gpm/ft² or more in some applications.

A formerly common arrangement for demineralizers was cation-anion-mixed bed ion exchange, although variations were not infrequent. Regardless, clarification does little or nothing to remove dissolved solids, so the TDS loading on the cation and anion beds was quite substantial. Resins will quickly exhaust in these conditions, and must be regenerated regularly with acid (typically sulfuric) and caustic.

For these reasons, makeup water treatment evolved considerably. A common scheme today is micro- or ultrafiltration (MF, UF) for suspended solids removal, two-pass RO for primary TDS removal, and portable mixed-bed ion exchange or electrodeionization (EDI) for final polishing. These technologies have greatly improved makeup system reliability and final water quality, but they are not foolproof. Issues that may arise include:

Excessive surges in suspended solids can foul MF and UF membranes.
Use of cationic polymers for coagulation or flocculation ahead of membrane systems may cause problems, as the polymers will coat the (typically) negatively-charged membranes.
For MF and UF membranes, the choice of outside-in vs. inside-out flow through the membranes may have significant consequences.
Membrane cleaning methods may induce scale formation depending upon the concentration of some impurities in the dilution water.
Although some vendors claim otherwise, in the author’s experience, EDI requires feed from a two-pass RO, not single-pass RO only.

Now let us examine a steam generation chemistry topic that continues to be misunderstood.

Forget the Oxygen Scavenger

Unlike coal-fired power plants, the condensate/feedwater system of an HRSG does not have feedwater heaters other than perhaps a deaerating heater. Thus, the entire network from the condenser to the low-pressure evaporator is all steel, with no copper alloys present. Even so, many HRSG proposals continue to call for oxygen scavenger feed to the condensate. This chemistry has been discredited as it is known that conditions produced by the reducing agent help to propagate and induce single-phase flow-accelerated corrosion (FAC).

Single-phase FAC occurs at flow disturbances such as elbows in feedwater piping and economizers, locations downstream of valves and reducing fittings, attemperator piping, and, most notably for the combined-cycle industry, in low-pressure evaporators. The effect of single-phase FAC is illustrated in the image.

Photo of tube-wall thinning caused by single-phase FAC. Image source: ChemTreat. Wall thinning occurs gradually until the remaining material at the affected location can no longer withstand the process pressure. Catastrophic failure may occur. Temperature and pH also influence single-phase FAC, with maximum corrosion occurring at 300^oF and as pH drops towards and below 9.0. In an HRSG, flow-accelerated corrosion is typically maximized in the unit’s low-pressure economizer and evaporator, with their many short-radius elbows and temperatures in proximity to 300^oF. However, the corrosion also can occur in other locations including intermediate-pressure circuits. Also, equipment such as deaerators that experience a mixture of water and steam may be subject to two-phase FAC.

Continued research is improving the battle against FAC. For those in the industry who have grasped that the old AVT(R) program (ammonia or amine feed for pH control and reducing agent feed for oxygen removal and metal passivation) is no longer valid, the popular replacement is all-volatile treatment (oxidizing). With AVT(O), the small amount of oxygen that enters via condenser air in-leakage is allowed to remain, and may be supplemented with a small, pure oxygen feed. Ammonia (or in some cases amine) is still ued to elevate the feedwater pH. At feedwater-specific conductivity below 0.2 µS/cm, this chemistry causes the magnetite layer on carbon steel to become interspersed and overlayed with a stronger, denser layer of ferric oxide hydrate (FeOOH), which essentially eliminates FAC in those locations. New guidelines recommend a dissolved oxygen concentration of 5 to 10 parts-per-billion at the LP economizer inlet as lower concentrations might allow gaps to form in the FeOOH coverage.

References