Toronto, ON, Canada
Marvin D. Silbert
Marvin Silbert and Associates
Willowdale, ON, Canada
The introduction of phosphate-based boiler-water alkalinity control brought a major improvement to the development of programs with pH control based upon free NaOH. The application of phosphate either by itself, or in conjunction with caustic soda, could provide the elevated pH necessary to minimize the corrosion of carbon steel. At the same time, the phosphate also provided sufficient chemical buffering capacity to cope with small system upsets, e.g., a small condenser leak. Phosphate was also economical and easy to use. While phosphate had the ability to significantly curtail incidents of caustic embrittlement and caustic gouging damage, it could not totally remove the fear of having free hydroxide in excess of that associated with trisodium phosphate. There was a need to get tighter control.
The introduction of Coordinated Phosphate Treatment or captive alkalinity program by Purcell and Whirl (1943) gave a significant improvement on the control of free hydroxide by placing an upper limit upon the pH (or alkalinity) allowed relative to the concentration of phosphate present. Coordinated phosphate treatment was simple process to monitor and control. To prevent the formation of free NaOH, the operator had to keep below the upper line in the plot of pH vs. phosphate concentration plot shown in Figure 1. That line corresponds to three sodiums for each phosphate, i.e., the Na/PO4 ratio of 3.0 found in pure trisodium phosphate.
Marcy and Halstead (1964) further reduced the risk of free NaOH with Congruent Phosphate Treatment (CPT). This modification controlled the Na/PO4 ratio below 2.6 while also maintaining the alkalinity high enough to minimize corrosion of carbon steel. Various authors have added empirical corrections to further subdivide the CPT region into a variety of boxes corresponding to an individual boiler's operating pressure. A typical high-pressure boiler in a fossil-fired electrical generating station would tend to run with the phosphate concentration kept in the 2-5 mg/L range while a lower pressure nuclear PWR (Pressurized Water reactor) or plant utility boiler might go to 25 mg/L or higher.
While the introduction of CPT did improve the situation with respect to attack by free caustic, it did not eliminate failures. Failures have been reported most frequently with the high-pressure boilers used for electrical generation. Dick (1963) reported that the reduction of phosphate concentration on a unit experiencing severe hideout led to hydrogen damage in the water wall tubes. Similar problems occurred after modifying boiler water chemistry control in four Ontario Hydro boilers in the early 1970s. Hydrogen damage is a sign of local acidic environments in contact with the tube surface usually in the presence of a high chloride concentration. Layton ( 1987) and Jonas and Layton (1988) documented extensive corrosion attack by phosphate in the Hunter No. 3 boiler of Utah Power and Light Company where a major boiler retubing rehabilitation was needed after only 18 months of operation. Goldstrohm and Robertson (1989) reported a similar, but much milder attack. Corrosion fatigue failures are also a major source of boiler unreliability currently investigated by EPRI (1993).
The damage to boilers as a result of the inadequacies of CPT treatment has not been restricted to the high-pressure boilers. It can also occur with plant utility and cogeneration boilers. Mort et al. (1995) reported tube damage in an 8.9 MPa (1300 psi) utility boiler system. The nuclear steam generators in PWR (pressurized water reactor) systems operate at pressures that are significantly lower than those used in most fossil-fired electrical generating plants. They also use tubing alloys that may be less-forgiving than carbon steel. Operators trying to meet CPT specifications ended up with PO4 concentrations exceeding 50 mg/L. With the high sensitivity of the tube alloys to caustic attack, there was an added effort in the PWR to keep the Na/PO4 ratio below 2.4 to avoid any likelihood of free caustic forming and damaging the tube materials. To get the ratio down, it was often necessary to blend disodium phosphate with the trisodium phosphate. In doing so, some plants operated with Na/PO4 ratios as low as 1.8. Many of these plants experienced a phenomenon known as phosphate wastage, where acidic phosphate attack resulted in severe wall thinning. Rather than find ways to better control their phosphate programs, the industry chose to switch to all volatile treatment (AVT) programs based upon hydrazine and ammonia/morpholine.
Why have these problems developed in plants that tried to rigidly keep to the CPT guidelines and why have many of the problems been acid attack rather than caustic attack? On closer examination, there is a major weakness in the monitoring and control of a CPT program. To achieve the final Na/PO4 ratio, it appeared to be necessary to adjust the chemistry, on a continual basis. The ratio could be increased by adding NaOH or TSP and decreased by adding DSP (disodium phosphate). This boiler water sample, upon which the control is based, represents an average of all the water within the boiler and as such it contains water subjected to both the most severe and the least severe conditions within the boiler. Boiler water in a high heat-flux zone is likely to concentrate due to evaporation. Local conditions may prevent adequate mixing with the bulk liquid phase and over time, a concentration gradient will develop between the bulk boiler water and the localized solution in contact with the magnetite surface. Localized concentrations for some species may be two or more orders of magnitude higher than in the bulk boiler water (Cohen et al., 1962). Under these conditions, species, such as NaOH and NaCl, with relatively high solubilities tend to stay in solution, while those of more limited solubility, such as those incorporating phosphate species, may exceed their solubility limits, and undergo precipitation or hideout. As it is the design and operating practices for each individual boiler that determine the extent of these localized effects, it is not possible for CPT to offer a single control regime that fits all boilers.
Essentially all reported cases of boiler tubes experiencing phosphate-induced corrosion have been associated with phosphate hideout. In spite of hide-out being a localized phenomenon, it may become visible in the bulk sample when the boiler changes load. As the boiler pressure is brought up to full load, hide-out will appear as a loss of phosphate from solution accompanied by an increase in pH and alkalinity. Typically, the phosphate concentration may drop below the levels specified for coordinated or congruent treatments while the pH goes above 9.7. Following a load reduction or shutdown, the phosphate may start to reappear with the pH dropping, possibly below 9.0.
Some of the assumptions that form the basis of CPT are too simplistic. As discussed above, there is the question of hideout in the hotter regions that must be considered. There is also the assumption of congruent precipitation of pure sodium phosphate salts, i.e., that the composition of solids in liquid and solid phases is the same, with no change in boiler water pH or Na/PO4 ratio. With the extensive carbon steel surfaces in a boiler, it can be expected that the various phosphate species are likely to react with any iron-containing corrosion products. As the reaction products precipitate, there will be accompanying changes to the alkalinity of the resulting liquid and solid phases. Economy et al. (1975) and Connor and Panson (1983) have demonstrated these addition processes in the laboratory.
The mechanism for the reaction of phosphate with metal oxides has been updated by Tremaine et al. (1992, 1993). Their work found that the reaction of phosphate with magnetite results in the formation of two products. One has the composition Na4FeOH(PO4)2·1/3NaOH. If the Na/PO4 ratio was less than 2.5 (as can be expected from trying to keep within the bounds of CPT by blending in disodium or monosodium salts) and the temperature was in the 320-360°C region, the other product was NaFePO4 or maricite, formed as a result of these acidic phosphate salts attacking both the magnetite and the base metal. Mort et al. (1995) found evidence for maricite in the deposits removed from failed tubes.
2Na2HPO4 + Fe + ½O2—> NaFePO4
+ Na3PO4 +H2O
3NaH2PO4 + Fe3O4—>
3NaFePO4 + ½O2 + 3H2O
When the Na/PO4 ratio was above 2.5 and the temperature 320°C, the other product was Na2.6Fe0.2PO4 rather than maricite. If the Na/PO4 ratio went above 2.7 and the temperature was raised to 360°C, H2 was formed as a result of a high local concentration of NaOH first dissolving the magnetite and then base metal. The presence of the hydrogen gas can be an indication that caustic gouging is occurring.
2NaOH + Fe —> Na2FeO2 + H2
The recent development of Equilibrium Phosphate Control (EPT) by Ontario Hydro (Stodola 1986, 1989, 1991) provided a more rational approach to phosphate chemistry. Few high-pressure boilers have been able to safely maintain the 25 mg/L (PO4) limits of a CPT regime. In fact, some can not tolerate 1 mg/L. Attempts to maintain an artificially high control range in such units will destabilize boiler water alkalinity control, increase the risks of dissolution of the protective magnetite layer and generate iron-phosphate based deposits. Forcing a boiler to operate with phosphate concentrations in excess of its equilibrium value means that any changes in the heat and mass transfer patterns in the high heat flux zone may result in corresponding changes in Na/PO4 ratios and the rates of precipitation and hydrolysis. Since solubility of phosphates increases progressively with a drop in Na/PO4 below 2.0, the addition of disodium or even monosodium phosphate for pH control could result in the formation of strongly acidic environments. A possible explanation of the incidents described by Dick and Stodola might be the production of phosphate generated acidity under deposits (e.g. within a dirty boiler) in the presence of significant concentrations of chloride with the tubes failing as a result of hydrogen production.
The fact that hydrogen production can result from both caustic gouging and acid phosphate attack, may have led to a number of incorrect diagnoses and, as a result, making the wrong response. If acid phosphate attack under congruent phosphate control is diagnosed as caustic gouging, the obvious response would be to further reduce the presence of free caustic by adding more disodium or monosodium phosphate. If the actual problem resulted from acidic phosphates, such a reaction would aggravate rather than help the situation.
Additions of sodium phosphates of higher Na/PO4 ratios (such as in coordinated phosphate treatment), in excess of equilibrium concentration results in extra quantities of solid phosphates and free caustic. The additional caustic, generated by the phosphate hideout mechanism, may further increase the potential for high-pH related corrosion in marginally designed boilers or those running above normal maximum load.
Phosphate is also known to affect crystalline structure of the adherent magnetite and may possibly alter its protective properties. Phosphate species become incorporated in the magnetite lattice, apparently causing weak spots (Stodola 1991, Kirsch 1964, Broadbent 1978). For this reason, Ontario Hydro avoids the use of trisodium phosphate in the passivation-neutralization step of the standard HCl boiler clean. To ensure good quality adherent magnetite coating, only NaOH is used to alkalize the boiler water during the first 500 operating hours following the clean. The level of contamination should be maintained as a minimum within EPRI Guidelines (MacDonald 1988), or better yet, within tighter pressure dependent limits recommended in Ontario Hydro specifications (EPRI 1986).
For many years, the CPT control chart has been accepted by the plant operator as a simple tool to maintain the system within a set of pre-established specifications. As discussed above, the ability to control the system requires a knowledge of the concentration at the tube surface. The ability to sample at this point is not possible; therefore, it is necessary to deduce whether or not deposition is occurring from analysis of the bulk fluid.
a. If the concentration of phosphate is in equilibrium with the system, then the phosphate should remain in solution with blowdown as the only mechanism by which it will be changed. If the phosphate addition is continuous, the bulk sample, will then represent the steady-state average from balancing what is fed into the boiler with what is being removed by blowdown. If the addition pump can be isolated and no more phosphate added, the phosphate concentration should fall at a rate dependent upon the blowdown flow. If there is no blowdown, then the phosphate concentration should remain constant. The PO4, pH and Na/PO4 ratio should also remain fairly constant in spite of load changes.
b. If phosphate is present at a concentration in excess of the equilibrium value, the excess will deposit within the system. This deposition process provides a second removal mechanism in addition to blowdown. The phosphate concentration within the bulk fluid will drop at a rate dependent upon the sum of the rates of blowdown and deposition until such time as the equilibrium concentration is reached. At that point, the deposition will cease and the concentration will fall only in proportion to the blowdown rate. While the PO4 is above the equilibrium value, the PO4, pH and Na/PO4 ratio are likely to show movement in response to load changes.
c. If the phosphate is present at a concentration below the equilibrium value, deposits from tube surfaces within the system will redissolve. The redissolution would be seen as a sudden increase in phosphate when a unit trips. It may also, if given enough time, provide a mechanism to at least partially remove old deposits. It also provides the necessary alkalinity to maintain the system during a shutdown state.
Initiating and maintaining an EPT program requires determining what concentration can be tolerated before deposition occurs and keeping at or below that value. It is a simple and straightforward procedure to determine where the equilibrium value sits. Add an excess of phosphate to bring the concentration to a value somewhat higher than the anticipated equilibrium value. Stop the addition and follow the decay of the phosphate in the bulk boiler sample. When the phosphate concentration is above the equilibrium value, hideout will occur and the phosphate concentration will drop and continue to do so until the equilibrium value is reached. The maximum operating range is the value at which it levels off. The PO4, pH and Na/PO4 ratio will remain stable with load changes, if the plant is operated within that range.
A number of boilers have been converted to EPT. They show equilibrium phosphate concentrations that are lower than those commonly accepted for CPT. For boilers operating at 16-19 MPa (2400-2800 psi), typical equilibrium concentrations have been less than 2 mg/L and sometimes less than 1 mg/L . Stodola (1986) found an upper limit of 2.4 mg/L for the Ontario Hydro units. Carter and Selby (1992) determined that the conditions for the E.D. Edwards Unit #2 required the limit to be less than 0.15 mg/L. Goldstrohm and Robertson (1989) found the equilibrium value for the 2600 psi Salt River Coronado station to be 0.7 mg/L. In each case, the change has better enabled the boilers to cope with transients. Under the earlier CPT, the boilers experienced very significant shifts with load swings. With the reduced phosphate concentrations of an EPT program, the shifts were essentially eliminated.
The examples in the above paragraph have demonstrated that operation of a
fossil-fired boiler within the EPT region can be a simple and straightforward
process and that doing so results in stable system chemistry with respect to
system upsets and also avoids problems with the excess NaOH or with acidic
phosphates inherent within the CPT control regime. While the results may not be
as clear with the nuclear PWRs, Silbert and MacNeil (1990) showed that it was
possible to operate within a range that could minimize the wastage problems
characteristic of those plants operating under CPT.
Experience shows that bulk boiler water alkalinity control, within a stable pH range of 9.0 to 9.7, is necessary for development of a good protective layer. As the plants that experience phosphate-related problems have used tri- and disodium phosphate blends to keep within the limits imposed by the CPT, congruent treatment program, it is important to add only trisodium phosphate, with the possibility of adding some NaOH during shutdowns. There are a number of boilers, particularly in Russia (Vasilenko 1984) that have operated with the addition of tri-sodium phosphate alone. There are also a number of boilers in the UK (Ball 1984) that operate using caustic soda as the one and only additive in spite of the concerns regarding caustic attack.
Figure 2 places the EPT regime on the phosphate-pH curve. The range extends upward from the trisodium phosphate line into the free caustic region, i.e. the opposite direction from the CPT region. The operation of an EPT program is quite different from the operation of the more traditional CPT program. The latter was based upon selecting a target range and forcing the boiler to operate within it. The concentration of phosphate which one boiler can tolerate may differ quite significantly from that acceptable to another boiler. Forcing all boilers to operate within this single set of specifications may result in severe damage (Silbert and Miyamoto 1993) if the individual boiler is not be able to hold that quantity of phosphate in solution at those surfaces experiencing higher heat fluxes. If the concentration is too high, there is only one place for the excess to go. It will end up as depositing on the tube surfaces.
With EPT, it is important to recognize that there is no specification to which the boiler must be forced to perform. Dooley et al. (1994) have provided a roadmap showing the steps to make the change. They do not provide specifications as it is the operator's job to ascertain where the EPT range will be for the individual boiler. Rather than seek a position on a chart, the control is based upon establishing an on-going program to evaluate (and re-evaluate) a set of specifications specific to a single plant under one or more sets of operating conditions. It is also important to recognize that a continuous addition system is capable of masking any deposition that occurs and makes it impossible to determine, on an on-going basis, whether or not deposition is occurring.
It can be expected that phosphate treatment will continue to be the dominant boiler treatment regime with EPT replacing CPT. While most test programs have been conducted by major electrical utilities, it can be expected that the industrial plant utilities and cogeneration plants will also be moving this way. When an EPT program is initiated, the absence of a fixed and simplified set of externally imposed specifications is likely to introduce some very interesting implications with respect to training programs and operating policies and procedures. Once the program is in full operation, deposit inventory and chemical cleaning frequency may provide a reliable measure of its effectiveness. An EPRI (1984) survey found that the average time between boiler cleanings for American boilers in excess of 1800 psi averages 4 years. Application of EPT in Ontario Hydro resulted initially in an extension of periods between chemical cleans from every 25,000 operating hours (approximately 5 years) to every 50,000 operating hours (10 years) now. It is felt that cleaning intervals of 16 to 20 years will be possible in the future.
Ball, M. and Dooley, R.B., "Sodium hydroxide for conditioning the boiler
water of drum-type boilers", EPRI TR-104007, Electric Power Research
Institute, Palo Alto, CA, May 1984
Broadbent, D., Lewis, G.G., and Wetton, E., "The chemistry of high
temperature phosphate solutions in relation to steam generation", Water
Chemistry of Nuclear Reactor Systems, BNES, pp 53-62, 1978
Carter, P.B., and Selbey, K.A., "Conversion of a high pressure drum boiler
from AVT to equilibrium phosphate chemistry", Proceedings of 53rd
International Water Conference, pp 472-480, Pittsburgh, 1992
Cohen, P., Taylor, G.R. and Picone, L.F., "A radiotracer test facility for
hideout measurements under boiler conditions." ASME Paper No.
Connor, W.M. and Panson, A.J., "Investigation of phosphatesludge
interactions", EPRI NP2963. Electric Power Research Institute, Palo
Alto, CA, March 1983
Dick, I.B., "Experiences with hydrogen embrittlement in the Consolidated
Edison System", ASME Paper No. 63PWR13
Dooley, R.B., Aschoff, A. and Pocock, F., "Cycle Chemistry Guidelines for
fossil plants: phosphate treatment for drum units", EPRI TR103665,
Electric Power Research Institute, Palo Alto, CA, December 1994
Economy, G., Panson, A.J., ChiaTsun Liu, Esposito, J.N., and Lindsay,
W.T. Jr, "Sodium phosphate solutions at boiler conditions: solubility, phase
equilibria, and interactions with magnetite", Proceedings of 35th
International Water Conference, pp 161173, Pittsburgh, 1975
EPRI, "Manual on Chemical Cleaning of FossilFuelled Steam Generation
Equipment", EPRI CS3289, pp 630, Electric Power Research
Institute, Palo Alto, CA, January 1984
EPRI, "Guidelines on fossil plant cycle chemistry", EPRI CS 4629,
Electric Power Research Institute, Palo Alto, CA, June 1986
EPRI, "Corrosion fatigue boiler tube failures in waterwalls and economizers",
EPRI TR-100455, Palo Alto, California, Electric Power Research Institute,
Palo Alto, April 1992-December 1993
Goldstrohm, D.D. and Robertson, T.W., "Low phosphate low sodium
hydroxide treatment in 2600 psig boilers of Salt River Project's Coronado
Station." Proceedings of 50th International Water Conference, pp
493499, Pittsburgh, 1989
Jonas, O. and Layton, K.F., "Phosphate boiler water treatment for high
pressure boilers", Proceedings of Second Fossil Plant Cycle Chemistry
Conference, Seattle, GS-6166, August 1988
Kirsch, H., "Phosphate containing deposits in high pressure steam
generators", Mitteilung VGB 89, PP80-88, 1964
Layton, K.F., "Water side corrosion in the water wall tubes of Hunter Unit
3", Proceedings of EPRI Conference on Boiler Tube Failures in Fossil Plants,
EPRI-1890, pp 2-271 to 2-286, November 1987
Marcy, V.M. and Halstead, S.L., "Improved basis for coordinated phosphate
pH control of boiler water." Combustion 45, pp 4547, 1964
McDonald, A.G., and Tracy, B.L., Phosphate hide-out in high pressure
solutions", Proceedings of 49th International Water Conference, pp
404-410, Pittsburgh, 1988
Mort, T., Rogers, M.E., Skinner, B., and Parson, R., "CO boiler reliability:
examination of failure mechanisms, root causes, corrective actions and
monitoring required", Proceedings of NACE Corrosion 95, Paper #620,
Purcell,T.E. and Whirl, S.F., "Protection against caustic embrittlement by
coordinated phosphate pH control", Transactions of The Electrochemical
Society 83, pp 279295, 1943
Silbert, M.D., and MacNeil, C.K., "Point Lepreau 600 MW CANDU experience",
American Society of Mechanical Engineers (ASME) Research and Technology
Committee on Water in Thermal Power Systems Workshop on Lab and Field
Experiences with Phosphates in Power Utility Drum Boilers, Pittsburgh, Oct
Silbert, M.D., and Miyamoto, H.K., "Reassessment of the role of
specifications in plant control philosophy", Industrial Water Treatment, (10
#5), pp 30-34, 1993
Stodola, J., "Review of boiler water alkalinity control." Proceedings of
46th International Water Conference, pp 235244, Pittsburgh, 1986
Stodola, J., "Boiler water chemistry control in Ontario Hydro high pressure
boilers", Proceedings of Second Fossil Plant Cycle Chemistry Conference. EPRI
GS6166. Electric Power Research Institute, Palo Alto, CA, January
Stodola, J., "Rationale for equilibrium phosphate treatment." Proceedings
of International Conference on Fossil Plant Cycle Chemistry, Baltimore, EPRI
TR100195, Electric Power Research Institute, Palo Alto, CA, 1991
Tremaine, P.R., Gray, L. and Stodola, J., "Sodiumphosphate chemistry
under highpressure utility drum boiler conditions", Canadian Electrical
Association Report 913 G 730, 1992
Tremaine, P.R., Gray, L.G.S., Wiwhar, B., Taylor, P. and Stodola, J.,
Phosphate interactions with metal oxides under high-pressure conditions",
Proceedings of 54th International Water Conference, pp 186-196,
Vasilenko, G.V., Sutotsky, G.P. and Dooley, R.B., "Investigation and
operating experience of Russian high pressure drum boilers using phosphate
treatment", EPRI TR104008, Electric Power Research Institute, Palo
Alto, CA, April 1984
Return to Corrosionsource.com Home Page