- Email: [email protected]
Flow-accelerated corrosion of pressure vessels in fossil plants
International Journal of Pressure Vessels and Piping 77 (2000) 85±90
www.elsevier.com/locate/ijpvp
Flow-accelerated corrosion of pressure vessels in fossil plants R.B. Dooley*, V.K. Chexal EPRI 3412 Hillview Avenue, Palo Alto, CA 94304, USA
Abstract Flow-accelerated corrosion (FAC) is a phenomenon that results in metal loss from piping, vessels and equipment made of carbon steel. It occurs under conditions of ¯ow, geometry and material, which are common in high-energy piping and tubing in nuclear, fossil and industrial power plants. Substantial progress has been made towards understanding the mechanism and in preventing FAC. This paper provides a sprinkling of that knowledge with particular emphasis for fossil and industrial plants. q 2000 Elsevier Science Ltd. All rights reserved. Keywords: Flow-accelerated corrosion; Feedwater; Hydrazine
1. Introduction Flow-accelerated corrosion (FAC) causes wall thinning (metal loss) of carbon steel piping, tubing and vessels exposed to ¯owing water or wet steam. If undetected, the degraded component can suddenly rupture, releasing high temperature steam and water into neighboring plant areas. The escaping ¯uids can injure plant workers, sometimes severely, and damage nearby equipment. Over the years, FAC has caused hundreds of piping and equipment failures in all types of fossil, industrial steam, and nuclear power plants. However, often the cause of the failure was not known by the plant owner, or if known, was not reported. Additionally, the power industry did not fully understand the conditions under which FAC occurred, where plants should look to ®nd it, or how to best control it when it was found. This changed in 1986. On 9th December of that year, an elbow in the condensate system ruptured at the Surry Nuclear Power Station. The failure caused four fatalities and tens of millions of dollars in repair costs and lost revenue. FAC was found to be the cause of the failure. Because of the deaths involved and the high degree of regulation applied to the nuclear power plants, a comprehensive overall approach was needed. An intensive international cooperative effort was initiated to understand the parameters which affect FAC. The strategy was that understanding FAC would allow the development of technology to help plants ®nd damage before failure occurs, and the measures to control it. The major parties in this cooperation were
* Corresponding author.
EPRI, Electricite de France, and Kraftwerk Union, now a part of Siemens. Flow-accelerated corrosion is a process whereby the normally protective oxide layer on carbon or low-alloy steel dissolves into a stream of ¯owing water or a water± steam mixture. The oxide layer becomes thinner and less protective, and the corrosion rate increases. Eventually a steady state is reached were the corrosion and dissolution rates are equal and stable corrosion rates are maintained. In some areas, the oxide layer may be so thin as to expose an apparently bare metal surface. More commonly, however, the corroded surface exhibits a black color typical of magnetite. Damage caused by ¯ow-accelerated corrosion can be characterized as a general reduction of wall thickness rather than a local attack, such as pitting or cracking. Although FAC occurs over a wide area within a given ®tting, it is localized in the sense that it frequently occurs over a limited area of piping ®tting due to local high areas of turbulence. In this content, ªlocalizedº may mean within several feet (,1 m) of the ®tting or region of turbulence. However, if one ®tting is found to be thinned, then most likely there will be others that have also lost material. A thinned component will typically fail due to overstress from operating pressure, or abrupt changes in conditions such as water hammer, and start-up loading. Large ®ttings may rupture suddenly rather than provide warning of their degraded condition by ®rst leaking. FAC occurs under both single and two-phase ¯ow conditions. Because water is necessary in order to remove the oxide layer, FAC does not occur in lines transporting dry or superheated steam.
0308-0161/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved. PII: S 0308-016 1(99)00087-3
86
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
Two-phase FAC has been recognized as a worldwide problem since about 1970. Since the mid-1980s, singlephase FAC has been acknowledged as a major problem in the balance-of-plant and secondary piping of US and foreign nuclear and fossil plants. The remainder of this paper concentrates only on single-phase FAC. 1.1. Some key ®eld history The technology and information developed since the Surry failure have greatly reduced the incidence of FAC failures, particularly in the nuclear plants which have the greatest level of implementation of the technology. Nevertheless, instances of severe thinning, leaks, and ruptures have still occurred. Some of the most signi®cant examples of recent failures in nuclear plants are summarized in Table 1 [1]. Failures in fossil and industrial steam plants have not historically been as well documented, because the plants are not as tightly regulated and because FAC has not been properly identi®ed. However, as a result of recent fatalities at a fossil plant, the topic now has a higher priority. A recent survey [2] of 63 utilities ranked the most important areas where FAC has been identi®ed in fossil plants (Table 2); 40% of the utilities had found FAC. Table 3 provides a summary of recent (from 1982) serious fossil plant failures, resulting in bursts, extensive plant damage, or fatalities, known to be caused by FAC. Table 3 indicates that there are a number of very important and signi®cant features which are common to these FAC incidents: ² The feedwater failures all occurred in the high-pressure portions of the system up to and including the economizer inlet header tubes. This means that the temperature range up to 280±3008C in fossil plants is susceptible to failure. This range is above the temperature at which maximum FAC occurs in laboratory experiments and above the temperature where magnetite solubility is maximum. ² Each system (A±W) had stainless steel low- and highpressure feedwater heaters. ² The feedwater in each case was treated with both ammonia and a reducing agent (usually hydrazine) which means the feedwater was operating under Table 1 Signi®cant FAC failures in nuclear power plants since 1989 Plant
Date
Location
S.M. de Garona (Spain) Loviisa Unit 1 (Finland) Millstone Unit 3 Millstone Unit 2 Almaraz Unit 1 (Spain) Loviisa Unit 2 (Finland) Sequoyah Unit 2 Fort Calhoun
December 1989 May 1990 December 1990 November 1991 December 1991 February 1993 March 1993 April 1997
Feedwater Feedwater Heater drain Reheater drain Extraction steam Feedwater Extraction steam Extraction steam
Table 2 Most important fossil plant areas experiencing FAC (results from 1997 survey of 63 utilities [2]) Piping around BFP Tubesheet/tubes in HP Heaters Heater Drain Lines Economizer Inlet Header Tubing Piping to Economizer Header Tubesheet/tubes in LP Heaters Deaerator shell Heat recovery steam generator (HRSG) tubing
reducing conditions (below 2300 mV ORP or redox potential). ² The feedwater oxygen levels were very low (,1 ppb). ² The heater drains are susceptible areas in plants with both all-ferrous and mixed feedwater metallurgies. FAC has also been commonly observed in industrial, pulp and paper, and chemical and petroleum plants. Most recently it has quickly become the most important failure mechanism in the heat recovery steam generators (HRSG) of combined cycle units. 1.2. Key features The FAC process produces a different surface appearance for single-phase ¯ow than it does for two-phase ¯ow: ² For single-phase ¯ow (liquid only): when the corrosion rate is high, the metal surface is often characterized by overlapping ªhorseshoe pitsº that give a scalloped or orange peel appearance. The scalloped appearance usually occurs in large diameter piping where a signi®cant loss of thickness has occurred. ² For two-phase ¯ow, the well-known ªtiger stripedº appearance is often observed in large pipes. The black part of the surface is the corroded area. It is covered with a very thin oxide ®lm. The blue to red colored part of the surface is protected by a thick oxide ®lm. This appearance is thought to be caused by a very disturbed, turbulent water ®lm ¯ow along the pipe wall. When FAC is occurring, the oxide ®lm present on the surface can be very thin (less than 1 mm). Under these conditions, the ®lm is nearly transparent and the surface has a metallic appearance. However, if the ®lm thickness increases, the surface appears black. 2. Mechanism of FAC The phenomenon of FAC is well-understood [1]. It is a process whereby the normally protective magnetite (Fe3O4) layer on carbon steel dissolves in a stream of ¯owing water or wet steam. This process reduces or eliminates the oxide layer and leads to a rapid removal of the base material until, in the worst cases, the pipe or tube bursts. The FAC process
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
87
Table 3 Serious FAC failures in fossil plants from 1982 [3] Plant
Location
Temp.
Feedwater treatment
A
Elbow downstream of BF booster pump Elbow near El at RT plug Downstream of boiler stop valve near El El tubes Drain lines
3608F 1828C
NH3/N2H4/Carbo-hydrazide
4568F 2368C
NH3/N2H4 NH3/N2H4
Final feedwater
NH3/N2H4
B C D±W Numerous
can become rapid: wall thinning rates as high as 3 mm/year have occurred. The rate of metal loss depends on a complex interplay of many parameters including the feedwater chemistry, the material composition, other materials in the feedwater systems, and the ¯uid hydrodynamics. The FAC process is an extension of the generalized corrosion process of carbon steel in stagnant water. The major difference is the effect of the water ¯ow at the oxide±solution interface. FAC can be divided into coupled processes that take into account the presence of the porous magnetite layer on the steel surface up to about 3008C (Fig. 1): ² The ®rst process produces soluble ferrous ions at the oxide±water interface and can be separated into three simultaneous actions: ± metal oxidation occurs at the iron±magnetite interface in water with a reducing potential (ORP , 0 mV); ± the ferrous species diffuse from the iron surface to the main water ¯ow through the porous oxide layer; ± the magnetite oxide layer at the oxide±water interface dissolves by a reductive process that is promoted by the presence of hydrogen. ² The second process involves the transfer of the ferrous ions into the bulk water across the boundary layer. The concentration of ferrous ions in the bulk water is very low compared to the concentration of ferrous ions at the oxide solution interface. The corrosion rate (FAC) increases if there is an increase of water ¯ow past the oxide±water interface.
Fig. 1. Schematic representation of oxide formed on iron-based feedwater surfaces during operation with deoxygenated all-volatile treatment (AVT) under reducing conditions (ORP p 0 mV).
pH
Oxygen
Feedwater heaters
Low , 1 ppb
All-stainless
8.80±9.20 8.75
Low , 1 ppb Very low , 1 ppb
All-stainless
9.00±9.40 Low
Very low , 1 ppb Very low , 1 ppb
All-stainless All-ferrous & mixed
This dissolution process is controlled by the oxidizing/ reducing potential (ORP) of the water. Basically the more reducing the feedwater, the higher is the dissolution and level of corrosion products measured in the feedwater. Under alkaline, deoxygenated (reducing) conditions the primary reaction of iron dissolution is inhibited by increasing pH or by decreasing the reducing environment (i.e. ORP becoming more positive). This causes a reduction of the ferrous ion (Fe 21 and Fe(OH) 1) concentration. The solubility of ferrous hydroxide (Fe(OH)2) rises with increasing temperature to a maximum at around 1508C, then decreases with a steep drop to the solubility of magnetite between 200 and 2508C. 2.1. Factors in¯uencing FAC The scienti®c basis of ¯ow-accelerated corrosion together with laboratory studies do not provide the utility engineer with the tools needed to solve real plant problems. To meet this need, a number of models have been developed to predict the rate of FAC [4±7]. There are several key parameters which affect the rate of FAC; these are discussed below. Because the FAC process as described above involves at least two steps (soluble iron production at the oxide/water interface and transfer of the corrosion products to the bulk solution across the diffusion boundary layer), it is necessary to consider both the corrosion processes (steel oxidation, oxide dissolution, charge transfer, etc.) and mass transfer. The role of the major in¯uencing factors can be classi®ed as follows: ² Hydrodynamics factors, i.e. ¯ow velocity, pipe roughness, geometry of the ¯ow path, steam quality or void fraction for two-phase ¯ows. ² Environmental factors, i.e. temperature, pH, reducing agent and oxygen concentration, oxidizing±reducing potential, water impurities. ² Metallurgical factors, mainly the chemical composition of the steel. The most bene®cial element has been determined to be chromium. An alloy with a nominal chromium content as low as 1% will have low or negligible FAC rates. There is evidence that even lower amounts of chromium (as low as 0.1%) may signi®cantly reduce FAC in single-phase systems.
88
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
Fig. 2. Fossil Plant FAC program Road Map.
2.2. Effect of water chemistry and oxidizing reducing potential It has been frequently observed in power plants, especially fossil and industrial plants, that the occurrence of FAC damage is very dependent on the environmental factors: (a) pH at temperature (b) ORP which provides an indication of the balance between reducing agents (such as hydrazine) and oxygen. Knowledge of the level of one or the other may not be suf®cient to indicate the susceptibility to FAC. Thus ORP is becoming the indicator of choice. It should be noted that ORP, also called redox, utilizes a platinum measuring electrode with a Ag/AgCl reference electrode to measure the potential. This measurement is made at ambient temperature with a sampling system in a similar manner to pH and dissolved oxygen. ORP is not a measurement of corrosion potential, but is an excellent parameter to determine the relative corrosion potential for a feedwater system [12]. Hydrazine (and alternatives) is a reducing agent added to the feedwater/condensate system in a power plant. It maintains a reducing environment in the steam generators (nuclear plants) and in the feedtrain. Hydrazine is unique in the chemical species in that it is reactive and unstable. It reacts with oxygen forming water and nitrogen. Most of the
hydrazine which does not react with oxygen thermally decomposes to form ammonia. Recent information indicates that in the 0±150 ppb range of hydrazine level, the FAC rate increases with increasing hydrazine level as the oxidizing± reducing potential (ORP) becomes more reducing. The decrease in potential in this range leads to greater dissolution of the surface magnetite (Fe3O4) and thus to an increase in the rate of FAC. Above the 150 ppb hydrazine level, the potential is lowered signi®cantly enough that it leads to slower kinetics. Thus, any further increase in the hydrazine level leads to a decrease in the FAC rate. Therefore, a plot of FAC rate versus hydrazine level is a bell-shaped curve with a peak at 150 ppb. Theoretical considerations show that FAC should be proportional to the concentration of hydrazine to the 1/6 power, with recent laboratory data indicating that this does not continue above 150 ppb of hydrazine or below 20 ppb. Hydrazine is commonly added to the feedwater of the PWR secondary circuit to keep feedwater oxygen levels lower than 5 ppb. Hydrazine is used to maintain a reducing environment in the feedtrain and the steam generator as a scavenger of residual oxygen and as a pH-conditioning agent as it decomposes to ammonia. However, to achieve a pH level of 9.5, a concentration of about 50 ppm is necessary, which is not feasible for plant applications. In fossil and industrial plants, hydrazine, if used, is added to the feedwater immediately after the condensate polishers or the condensate extraction pump. It is an absolute necessity for mixed-metallurgy systems to protect the copperbased feedwater heaters. However, in all-ferrous systems, it can be eliminated in high purity ,0:3mS=cm feedwater. In both cases, it is not optimum practice to use the hydrazine to both alkalize the cycle and to deoxygenate. In the past, hydrazine concentrations of about 20 ppb were used to control the oxygen in the feedtrain. Recently, US utilities have begun adding larger amounts of hydrazine in nuclear plants to maintain a strongly reducing environment within the steam generators. Hydrazine concentrations of approximately 100 ppb are used for this purpose; however some utilities, outside the US have used levels up to 600 ppb. (c) Temperature: Temperature is an important variable affecting the FAC of carbon steels and low alloyed steels. The phenomenon usually occurs between 100 and 2808C. Temperature in¯uences the rates of the oxidation and reduction reactions. It also in¯uences the values of the thermophysical properties. FAC as a function of temperature is a dome-shaped curve with a peak near 1508C for a constant mass ¯ow rate, steam quality, and water chemistry. Above and below the peak temperature, FAC wear rates tend to decrease. Note, however, that the peak temperature is a function of the main variables affecting the FAC process; but in practice the major failures have occurred in fossil and industrial plants at temperatures above this laboratory derived maximum (Table 3).
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
89
² Type C is oxygenated treatment (OT) using only ammonia and oxygen.
Fig. 3. Schematic representation of oxide formed on iron-based feedwater surfaces during operation with oxidizing feedwater.
3. Application of mechanism to FAC control A comprehensive FAC program can help a utility/operator avoid the large costs that accompany pipe/tube failure and forced outages. The aspects of such programs for nuclear [8] and fossil/industrial [9] plants have been developed. For example, Fig. 2 shows the two pronged approach for controlling FAC in fossil plants. The following are some of the activities in any FAC inspection, evaluation, and mitigation program: ² ² ² ² ² ² ² ² ² ²
identifying susceptible systems developing computer models of susceptible lines [4±7] selecting locations for inspection inspection planning [10] preparing for inspection conducting inspections [11] evaluating inspection data qualifying the components for continued use repair or replacing thinned components optimizing feedwater chemistry.
3.1. Control of feedwater chemistry in fossil/industrial plants Unlike in most nuclear plants where the feedwater has to be reducing to protect the steam generator, the feedwater chemistry can be changed/optimized in some fossil/industrial plants. These cycle chemistry activities have the proven capability of reducing the generation of feedwater corrosion products, and reducing and nearly eliminating FAC depending on the system metallurgy and the feedwater chemistry adopted. For both all-ferrous and mixed-metallurgy feedwater systems, the feedwater treatment always needs to be allvolatile. Currently there are three types of treatment: ² Type A is classical all-volatile treatment (AVT) using ammonia and a reducing agent or oxygen scavenger (such as hydrazine). ² Type B is the same as Type A minus the reducing agent, called ªNew AVTº.
The major difference in terms of FAC between Type A, and Types B and C chemistries is that Type A provides a reducing environment (ORP , 0 mV), whereas Types B and C result in an oxidizing environment (ORP . 0 mV). Under the reducing conditions of Type A feedwater chemistry, the type of oxide (magnetite) which grows on carbon steel surfaces up to about 3008C is shown schematically in Fig. 1. For ªnormalº Type A chemistry, the ORP is typically less than 2300 mV, and the level of feedwater corrosion products (measured at the economizer inlet) will be less than 10 ppb. Most fossil plants can easily achieve 5 ppb; this situation is regarded as ªnormalº and not of major concern from an FAC viewpoint. For a unit with Type A chemistry and high rates of FAC, the ORP will also be less than 2300 mV; but because of the local ¯ow hydrodynamics, the total corrosion/dissolution process is faster. The result is more dissolution and oxide particle entrainment into the ¯ow. In the worst cases, FAC will be so fast that there is only a very thin layer of Fe3O4 on the surface. Importantly, the level of feedwater corrosion products (measured at the economizer inlet) can be much more than 10 ppb. Types B and C chemistries produce a completely different situation. Fig. 3 shows the typical oxide structure that forms on carbon steel surfaces up to about 280±3008C under oxidizing feedwater environments. Here, the surface is covered by a layer of ferric oxide hydrate (FeOOH), which also permeates down the pores of the magnetite. So under the conditions of B and C chemistries the ORP is greater than zero; conditions which favor the growth of FeOOH. This formation does two things: (i) it reduces the overall corrosion rate because the diffusion (or access) of oxygen to the base material is restricted (or reduced), and (ii) it reduces the solubility of the surface oxide layers. Thus, from an FAC perspective, this surface FeOOH layer dissolves much slower (at least two orders of magnitude slower) than magnetite into the ¯owing feedwater, under exactly the same hydrodynamic conditions that existed previously with Type A chemistry. The overall result is that the measured feedwater corrosion products can be much less than 1 ppb and FAC is minimal. There are essentially two types of all-ferrous systems: those containing only carbon steel in the tubing and piping, and those containing stainless steel tubing and carbon steel piping. As can be seen clearly in Table 3 most of the serious FAC failures in fossil plants have occurred when all the tubing (both LP and HP) is stainless steel, and the chemistry is Type A (normal AVT with hydrazine producing ORP , 0 mV). This implies that the reducing environment is more severe in the HP feedwater when the tubing is stainless as compared to when the tubing is carbon steel or copper-based.
90
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
Table 4 Comparison of normal cycle chemistry limits at the economizer inlet for AVT feedwater and oxygenated treatments (values in parentheses represent the achievable and desirable levels) Cycle chemistry parameter
AVT (mixed metallurgy)
AVT (all-ferrous)
Oxygenated treatment (OT) (all-ferrous)
PH
8.8±9.1
9.2±9.6
Ammonia, NH3 ppm
0.15±0.4
0.5±2.0
Cation conductivity (mS/cm) Fe, ppb Cu, ppb Oxygen, ppb
, 0.2 , 10 (,5) , 2 (,2) , 5 (,2)
, 0.2 (,0.15) , 5 (,2)
8.0±8.5 a 9.0±9.5 b 0.02±0.07 a 0.3±1.5 b , 0.15 (,0.1) , 5 (,1)
1±10
ORP c, mV
p0
0±80
a b c
30±150 a 30±50 b . 100
For once-through units. For drum units. Oxidizing±reducing potential (with respect to Pt electrode vs Ag/AgCl).
Mixed-metallurgy systems can only use Type A feedwater chemistry which maintains a reducing environment under all operating regimes to protect the copper based tubing. This means that the carbon steel interconnecting piping and the economizer inlet tubing must also be exposed to the same reducing environment. Observation of Table 3 indicates that no serious FAC failures have occurred in fossil plant mixed-metallurgy systems; however wall loss associated with FAC has been observed in fossil plants, and serious failures have occurred in nuclear plants with mixed-metallurgy systems; thus the carbon steel components in mixed-metallurgy feedwater systems must be subjected to the same rigorous FAC programs as for allferrous systems. Finally in this fossil/industrial plant example, Table 4 provides a comparison of the optimum feedwater chemistry for all-ferrous and mixed metallurgy systems. 4. Concluding remarks Although FAC remains a major problem in nuclear, fossil and industrial plants, solutions and approaches are available. The thorough understanding of the mechanism in feedwater systems which have all-ferrous metallurgy and mixed-metallurgy allows the proper application of inspection and cycle chemistry approaches. The recent studies involving the effects of oxygen and reducing agents have led to the use of ORP as a main chemistry controlling parameter. Several computer programs to predict FAC are available, and together with the optimum inspection techniques and cycle chemistry should provide the tools to implement
FAC programs which will ensure an adequate level of safety.
References [1] Chexal VK, Horowitz J, Dooley RB, et al. Flow-accelerated corrosion in power plants. EPRI TR-106611R1, July 1998. [2] Dooley RB, Mathews J, editors. Fifth International Cycle Chemistry Conference. EPRI TR-108469, October 1997. [3] Dooley RB, Mathews J. The current state of cycle chemistry for fossil plants. Fifth International Cycle Chemistry Conference. EPRI TR108469, October 1997. [4] CHECWORKS Computer Program User Guide. EPRI TR-103198P1, December 1997. [5] CHECUP, a CHECWORKS Application for FAC Evaluation of Fossil Power Plants User Guide. EPRI TR-107066, September 1996. [6] BRT-CICERO. Flow-accelerated Corrosion Software. EdF, December 1997. [7] WATHEC/DASY. Flow-accelerated Corrosion Software. Siemens/ KWU, 1991. [8] Chexal VK, Munson DP, Horowitz JS, Randall GA. Recommendations for an effective ¯ow-accelerated corrosion program. NSAC202L-R1, EPRI, November 1996. [9] Dooley RB, Tilley RM, Chexal VK, Horowitz JS, Munson DP. Guidelines for controlling ¯ow-accelerated corrosion in fossil plants. TR108859, EPRI, November 1997. [10] Chexal VK, Dooley RB, Munson DP, Tilley RM. Control of FAC in Fossil, Co-generation and industrial steam plants. International Water Conference 1997. Paper IWC-97-57. [11] Tilley RM, Dooley RB, Brett CR. Recent developments in managing FAC in feedwater pipework for fossil-®red power plants. Joint Power Generation Conference, 1998. [12] Filer S, Janick M. ORP provides versatile water treatment power engineering, vol. 102, #11, November 1998. p. 50±6.
www.elsevier.com/locate/ijpvp
Flow-accelerated corrosion of pressure vessels in fossil plants R.B. Dooley*, V.K. Chexal EPRI 3412 Hillview Avenue, Palo Alto, CA 94304, USA
Abstract Flow-accelerated corrosion (FAC) is a phenomenon that results in metal loss from piping, vessels and equipment made of carbon steel. It occurs under conditions of ¯ow, geometry and material, which are common in high-energy piping and tubing in nuclear, fossil and industrial power plants. Substantial progress has been made towards understanding the mechanism and in preventing FAC. This paper provides a sprinkling of that knowledge with particular emphasis for fossil and industrial plants. q 2000 Elsevier Science Ltd. All rights reserved. Keywords: Flow-accelerated corrosion; Feedwater; Hydrazine
1. Introduction Flow-accelerated corrosion (FAC) causes wall thinning (metal loss) of carbon steel piping, tubing and vessels exposed to ¯owing water or wet steam. If undetected, the degraded component can suddenly rupture, releasing high temperature steam and water into neighboring plant areas. The escaping ¯uids can injure plant workers, sometimes severely, and damage nearby equipment. Over the years, FAC has caused hundreds of piping and equipment failures in all types of fossil, industrial steam, and nuclear power plants. However, often the cause of the failure was not known by the plant owner, or if known, was not reported. Additionally, the power industry did not fully understand the conditions under which FAC occurred, where plants should look to ®nd it, or how to best control it when it was found. This changed in 1986. On 9th December of that year, an elbow in the condensate system ruptured at the Surry Nuclear Power Station. The failure caused four fatalities and tens of millions of dollars in repair costs and lost revenue. FAC was found to be the cause of the failure. Because of the deaths involved and the high degree of regulation applied to the nuclear power plants, a comprehensive overall approach was needed. An intensive international cooperative effort was initiated to understand the parameters which affect FAC. The strategy was that understanding FAC would allow the development of technology to help plants ®nd damage before failure occurs, and the measures to control it. The major parties in this cooperation were
* Corresponding author.
EPRI, Electricite de France, and Kraftwerk Union, now a part of Siemens. Flow-accelerated corrosion is a process whereby the normally protective oxide layer on carbon or low-alloy steel dissolves into a stream of ¯owing water or a water± steam mixture. The oxide layer becomes thinner and less protective, and the corrosion rate increases. Eventually a steady state is reached were the corrosion and dissolution rates are equal and stable corrosion rates are maintained. In some areas, the oxide layer may be so thin as to expose an apparently bare metal surface. More commonly, however, the corroded surface exhibits a black color typical of magnetite. Damage caused by ¯ow-accelerated corrosion can be characterized as a general reduction of wall thickness rather than a local attack, such as pitting or cracking. Although FAC occurs over a wide area within a given ®tting, it is localized in the sense that it frequently occurs over a limited area of piping ®tting due to local high areas of turbulence. In this content, ªlocalizedº may mean within several feet (,1 m) of the ®tting or region of turbulence. However, if one ®tting is found to be thinned, then most likely there will be others that have also lost material. A thinned component will typically fail due to overstress from operating pressure, or abrupt changes in conditions such as water hammer, and start-up loading. Large ®ttings may rupture suddenly rather than provide warning of their degraded condition by ®rst leaking. FAC occurs under both single and two-phase ¯ow conditions. Because water is necessary in order to remove the oxide layer, FAC does not occur in lines transporting dry or superheated steam.
0308-0161/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved. PII: S 0308-016 1(99)00087-3
86
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
Two-phase FAC has been recognized as a worldwide problem since about 1970. Since the mid-1980s, singlephase FAC has been acknowledged as a major problem in the balance-of-plant and secondary piping of US and foreign nuclear and fossil plants. The remainder of this paper concentrates only on single-phase FAC. 1.1. Some key ®eld history The technology and information developed since the Surry failure have greatly reduced the incidence of FAC failures, particularly in the nuclear plants which have the greatest level of implementation of the technology. Nevertheless, instances of severe thinning, leaks, and ruptures have still occurred. Some of the most signi®cant examples of recent failures in nuclear plants are summarized in Table 1 [1]. Failures in fossil and industrial steam plants have not historically been as well documented, because the plants are not as tightly regulated and because FAC has not been properly identi®ed. However, as a result of recent fatalities at a fossil plant, the topic now has a higher priority. A recent survey [2] of 63 utilities ranked the most important areas where FAC has been identi®ed in fossil plants (Table 2); 40% of the utilities had found FAC. Table 3 provides a summary of recent (from 1982) serious fossil plant failures, resulting in bursts, extensive plant damage, or fatalities, known to be caused by FAC. Table 3 indicates that there are a number of very important and signi®cant features which are common to these FAC incidents: ² The feedwater failures all occurred in the high-pressure portions of the system up to and including the economizer inlet header tubes. This means that the temperature range up to 280±3008C in fossil plants is susceptible to failure. This range is above the temperature at which maximum FAC occurs in laboratory experiments and above the temperature where magnetite solubility is maximum. ² Each system (A±W) had stainless steel low- and highpressure feedwater heaters. ² The feedwater in each case was treated with both ammonia and a reducing agent (usually hydrazine) which means the feedwater was operating under Table 1 Signi®cant FAC failures in nuclear power plants since 1989 Plant
Date
Location
S.M. de Garona (Spain) Loviisa Unit 1 (Finland) Millstone Unit 3 Millstone Unit 2 Almaraz Unit 1 (Spain) Loviisa Unit 2 (Finland) Sequoyah Unit 2 Fort Calhoun
December 1989 May 1990 December 1990 November 1991 December 1991 February 1993 March 1993 April 1997
Feedwater Feedwater Heater drain Reheater drain Extraction steam Feedwater Extraction steam Extraction steam
Table 2 Most important fossil plant areas experiencing FAC (results from 1997 survey of 63 utilities [2]) Piping around BFP Tubesheet/tubes in HP Heaters Heater Drain Lines Economizer Inlet Header Tubing Piping to Economizer Header Tubesheet/tubes in LP Heaters Deaerator shell Heat recovery steam generator (HRSG) tubing
reducing conditions (below 2300 mV ORP or redox potential). ² The feedwater oxygen levels were very low (,1 ppb). ² The heater drains are susceptible areas in plants with both all-ferrous and mixed feedwater metallurgies. FAC has also been commonly observed in industrial, pulp and paper, and chemical and petroleum plants. Most recently it has quickly become the most important failure mechanism in the heat recovery steam generators (HRSG) of combined cycle units. 1.2. Key features The FAC process produces a different surface appearance for single-phase ¯ow than it does for two-phase ¯ow: ² For single-phase ¯ow (liquid only): when the corrosion rate is high, the metal surface is often characterized by overlapping ªhorseshoe pitsº that give a scalloped or orange peel appearance. The scalloped appearance usually occurs in large diameter piping where a signi®cant loss of thickness has occurred. ² For two-phase ¯ow, the well-known ªtiger stripedº appearance is often observed in large pipes. The black part of the surface is the corroded area. It is covered with a very thin oxide ®lm. The blue to red colored part of the surface is protected by a thick oxide ®lm. This appearance is thought to be caused by a very disturbed, turbulent water ®lm ¯ow along the pipe wall. When FAC is occurring, the oxide ®lm present on the surface can be very thin (less than 1 mm). Under these conditions, the ®lm is nearly transparent and the surface has a metallic appearance. However, if the ®lm thickness increases, the surface appears black. 2. Mechanism of FAC The phenomenon of FAC is well-understood [1]. It is a process whereby the normally protective magnetite (Fe3O4) layer on carbon steel dissolves in a stream of ¯owing water or wet steam. This process reduces or eliminates the oxide layer and leads to a rapid removal of the base material until, in the worst cases, the pipe or tube bursts. The FAC process
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
87
Table 3 Serious FAC failures in fossil plants from 1982 [3] Plant
Location
Temp.
Feedwater treatment
A
Elbow downstream of BF booster pump Elbow near El at RT plug Downstream of boiler stop valve near El El tubes Drain lines
3608F 1828C
NH3/N2H4/Carbo-hydrazide
4568F 2368C
NH3/N2H4 NH3/N2H4
Final feedwater
NH3/N2H4
B C D±W Numerous
can become rapid: wall thinning rates as high as 3 mm/year have occurred. The rate of metal loss depends on a complex interplay of many parameters including the feedwater chemistry, the material composition, other materials in the feedwater systems, and the ¯uid hydrodynamics. The FAC process is an extension of the generalized corrosion process of carbon steel in stagnant water. The major difference is the effect of the water ¯ow at the oxide±solution interface. FAC can be divided into coupled processes that take into account the presence of the porous magnetite layer on the steel surface up to about 3008C (Fig. 1): ² The ®rst process produces soluble ferrous ions at the oxide±water interface and can be separated into three simultaneous actions: ± metal oxidation occurs at the iron±magnetite interface in water with a reducing potential (ORP , 0 mV); ± the ferrous species diffuse from the iron surface to the main water ¯ow through the porous oxide layer; ± the magnetite oxide layer at the oxide±water interface dissolves by a reductive process that is promoted by the presence of hydrogen. ² The second process involves the transfer of the ferrous ions into the bulk water across the boundary layer. The concentration of ferrous ions in the bulk water is very low compared to the concentration of ferrous ions at the oxide solution interface. The corrosion rate (FAC) increases if there is an increase of water ¯ow past the oxide±water interface.
Fig. 1. Schematic representation of oxide formed on iron-based feedwater surfaces during operation with deoxygenated all-volatile treatment (AVT) under reducing conditions (ORP p 0 mV).
pH
Oxygen
Feedwater heaters
Low , 1 ppb
All-stainless
8.80±9.20 8.75
Low , 1 ppb Very low , 1 ppb
All-stainless
9.00±9.40 Low
Very low , 1 ppb Very low , 1 ppb
All-stainless All-ferrous & mixed
This dissolution process is controlled by the oxidizing/ reducing potential (ORP) of the water. Basically the more reducing the feedwater, the higher is the dissolution and level of corrosion products measured in the feedwater. Under alkaline, deoxygenated (reducing) conditions the primary reaction of iron dissolution is inhibited by increasing pH or by decreasing the reducing environment (i.e. ORP becoming more positive). This causes a reduction of the ferrous ion (Fe 21 and Fe(OH) 1) concentration. The solubility of ferrous hydroxide (Fe(OH)2) rises with increasing temperature to a maximum at around 1508C, then decreases with a steep drop to the solubility of magnetite between 200 and 2508C. 2.1. Factors in¯uencing FAC The scienti®c basis of ¯ow-accelerated corrosion together with laboratory studies do not provide the utility engineer with the tools needed to solve real plant problems. To meet this need, a number of models have been developed to predict the rate of FAC [4±7]. There are several key parameters which affect the rate of FAC; these are discussed below. Because the FAC process as described above involves at least two steps (soluble iron production at the oxide/water interface and transfer of the corrosion products to the bulk solution across the diffusion boundary layer), it is necessary to consider both the corrosion processes (steel oxidation, oxide dissolution, charge transfer, etc.) and mass transfer. The role of the major in¯uencing factors can be classi®ed as follows: ² Hydrodynamics factors, i.e. ¯ow velocity, pipe roughness, geometry of the ¯ow path, steam quality or void fraction for two-phase ¯ows. ² Environmental factors, i.e. temperature, pH, reducing agent and oxygen concentration, oxidizing±reducing potential, water impurities. ² Metallurgical factors, mainly the chemical composition of the steel. The most bene®cial element has been determined to be chromium. An alloy with a nominal chromium content as low as 1% will have low or negligible FAC rates. There is evidence that even lower amounts of chromium (as low as 0.1%) may signi®cantly reduce FAC in single-phase systems.
88
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
Fig. 2. Fossil Plant FAC program Road Map.
2.2. Effect of water chemistry and oxidizing reducing potential It has been frequently observed in power plants, especially fossil and industrial plants, that the occurrence of FAC damage is very dependent on the environmental factors: (a) pH at temperature (b) ORP which provides an indication of the balance between reducing agents (such as hydrazine) and oxygen. Knowledge of the level of one or the other may not be suf®cient to indicate the susceptibility to FAC. Thus ORP is becoming the indicator of choice. It should be noted that ORP, also called redox, utilizes a platinum measuring electrode with a Ag/AgCl reference electrode to measure the potential. This measurement is made at ambient temperature with a sampling system in a similar manner to pH and dissolved oxygen. ORP is not a measurement of corrosion potential, but is an excellent parameter to determine the relative corrosion potential for a feedwater system [12]. Hydrazine (and alternatives) is a reducing agent added to the feedwater/condensate system in a power plant. It maintains a reducing environment in the steam generators (nuclear plants) and in the feedtrain. Hydrazine is unique in the chemical species in that it is reactive and unstable. It reacts with oxygen forming water and nitrogen. Most of the
hydrazine which does not react with oxygen thermally decomposes to form ammonia. Recent information indicates that in the 0±150 ppb range of hydrazine level, the FAC rate increases with increasing hydrazine level as the oxidizing± reducing potential (ORP) becomes more reducing. The decrease in potential in this range leads to greater dissolution of the surface magnetite (Fe3O4) and thus to an increase in the rate of FAC. Above the 150 ppb hydrazine level, the potential is lowered signi®cantly enough that it leads to slower kinetics. Thus, any further increase in the hydrazine level leads to a decrease in the FAC rate. Therefore, a plot of FAC rate versus hydrazine level is a bell-shaped curve with a peak at 150 ppb. Theoretical considerations show that FAC should be proportional to the concentration of hydrazine to the 1/6 power, with recent laboratory data indicating that this does not continue above 150 ppb of hydrazine or below 20 ppb. Hydrazine is commonly added to the feedwater of the PWR secondary circuit to keep feedwater oxygen levels lower than 5 ppb. Hydrazine is used to maintain a reducing environment in the feedtrain and the steam generator as a scavenger of residual oxygen and as a pH-conditioning agent as it decomposes to ammonia. However, to achieve a pH level of 9.5, a concentration of about 50 ppm is necessary, which is not feasible for plant applications. In fossil and industrial plants, hydrazine, if used, is added to the feedwater immediately after the condensate polishers or the condensate extraction pump. It is an absolute necessity for mixed-metallurgy systems to protect the copperbased feedwater heaters. However, in all-ferrous systems, it can be eliminated in high purity ,0:3mS=cm feedwater. In both cases, it is not optimum practice to use the hydrazine to both alkalize the cycle and to deoxygenate. In the past, hydrazine concentrations of about 20 ppb were used to control the oxygen in the feedtrain. Recently, US utilities have begun adding larger amounts of hydrazine in nuclear plants to maintain a strongly reducing environment within the steam generators. Hydrazine concentrations of approximately 100 ppb are used for this purpose; however some utilities, outside the US have used levels up to 600 ppb. (c) Temperature: Temperature is an important variable affecting the FAC of carbon steels and low alloyed steels. The phenomenon usually occurs between 100 and 2808C. Temperature in¯uences the rates of the oxidation and reduction reactions. It also in¯uences the values of the thermophysical properties. FAC as a function of temperature is a dome-shaped curve with a peak near 1508C for a constant mass ¯ow rate, steam quality, and water chemistry. Above and below the peak temperature, FAC wear rates tend to decrease. Note, however, that the peak temperature is a function of the main variables affecting the FAC process; but in practice the major failures have occurred in fossil and industrial plants at temperatures above this laboratory derived maximum (Table 3).
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
89
² Type C is oxygenated treatment (OT) using only ammonia and oxygen.
Fig. 3. Schematic representation of oxide formed on iron-based feedwater surfaces during operation with oxidizing feedwater.
3. Application of mechanism to FAC control A comprehensive FAC program can help a utility/operator avoid the large costs that accompany pipe/tube failure and forced outages. The aspects of such programs for nuclear [8] and fossil/industrial [9] plants have been developed. For example, Fig. 2 shows the two pronged approach for controlling FAC in fossil plants. The following are some of the activities in any FAC inspection, evaluation, and mitigation program: ² ² ² ² ² ² ² ² ² ²
identifying susceptible systems developing computer models of susceptible lines [4±7] selecting locations for inspection inspection planning [10] preparing for inspection conducting inspections [11] evaluating inspection data qualifying the components for continued use repair or replacing thinned components optimizing feedwater chemistry.
3.1. Control of feedwater chemistry in fossil/industrial plants Unlike in most nuclear plants where the feedwater has to be reducing to protect the steam generator, the feedwater chemistry can be changed/optimized in some fossil/industrial plants. These cycle chemistry activities have the proven capability of reducing the generation of feedwater corrosion products, and reducing and nearly eliminating FAC depending on the system metallurgy and the feedwater chemistry adopted. For both all-ferrous and mixed-metallurgy feedwater systems, the feedwater treatment always needs to be allvolatile. Currently there are three types of treatment: ² Type A is classical all-volatile treatment (AVT) using ammonia and a reducing agent or oxygen scavenger (such as hydrazine). ² Type B is the same as Type A minus the reducing agent, called ªNew AVTº.
The major difference in terms of FAC between Type A, and Types B and C chemistries is that Type A provides a reducing environment (ORP , 0 mV), whereas Types B and C result in an oxidizing environment (ORP . 0 mV). Under the reducing conditions of Type A feedwater chemistry, the type of oxide (magnetite) which grows on carbon steel surfaces up to about 3008C is shown schematically in Fig. 1. For ªnormalº Type A chemistry, the ORP is typically less than 2300 mV, and the level of feedwater corrosion products (measured at the economizer inlet) will be less than 10 ppb. Most fossil plants can easily achieve 5 ppb; this situation is regarded as ªnormalº and not of major concern from an FAC viewpoint. For a unit with Type A chemistry and high rates of FAC, the ORP will also be less than 2300 mV; but because of the local ¯ow hydrodynamics, the total corrosion/dissolution process is faster. The result is more dissolution and oxide particle entrainment into the ¯ow. In the worst cases, FAC will be so fast that there is only a very thin layer of Fe3O4 on the surface. Importantly, the level of feedwater corrosion products (measured at the economizer inlet) can be much more than 10 ppb. Types B and C chemistries produce a completely different situation. Fig. 3 shows the typical oxide structure that forms on carbon steel surfaces up to about 280±3008C under oxidizing feedwater environments. Here, the surface is covered by a layer of ferric oxide hydrate (FeOOH), which also permeates down the pores of the magnetite. So under the conditions of B and C chemistries the ORP is greater than zero; conditions which favor the growth of FeOOH. This formation does two things: (i) it reduces the overall corrosion rate because the diffusion (or access) of oxygen to the base material is restricted (or reduced), and (ii) it reduces the solubility of the surface oxide layers. Thus, from an FAC perspective, this surface FeOOH layer dissolves much slower (at least two orders of magnitude slower) than magnetite into the ¯owing feedwater, under exactly the same hydrodynamic conditions that existed previously with Type A chemistry. The overall result is that the measured feedwater corrosion products can be much less than 1 ppb and FAC is minimal. There are essentially two types of all-ferrous systems: those containing only carbon steel in the tubing and piping, and those containing stainless steel tubing and carbon steel piping. As can be seen clearly in Table 3 most of the serious FAC failures in fossil plants have occurred when all the tubing (both LP and HP) is stainless steel, and the chemistry is Type A (normal AVT with hydrazine producing ORP , 0 mV). This implies that the reducing environment is more severe in the HP feedwater when the tubing is stainless as compared to when the tubing is carbon steel or copper-based.
90
R.B. Dooley, V.K. Chexal / International Journal of Pressure Vessels and Piping 77 (2000) 85±90
Table 4 Comparison of normal cycle chemistry limits at the economizer inlet for AVT feedwater and oxygenated treatments (values in parentheses represent the achievable and desirable levels) Cycle chemistry parameter
AVT (mixed metallurgy)
AVT (all-ferrous)
Oxygenated treatment (OT) (all-ferrous)
PH
8.8±9.1
9.2±9.6
Ammonia, NH3 ppm
0.15±0.4
0.5±2.0
Cation conductivity (mS/cm) Fe, ppb Cu, ppb Oxygen, ppb
, 0.2 , 10 (,5) , 2 (,2) , 5 (,2)
, 0.2 (,0.15) , 5 (,2)
8.0±8.5 a 9.0±9.5 b 0.02±0.07 a 0.3±1.5 b , 0.15 (,0.1) , 5 (,1)
1±10
ORP c, mV
p0
0±80
a b c
30±150 a 30±50 b . 100
For once-through units. For drum units. Oxidizing±reducing potential (with respect to Pt electrode vs Ag/AgCl).
Mixed-metallurgy systems can only use Type A feedwater chemistry which maintains a reducing environment under all operating regimes to protect the copper based tubing. This means that the carbon steel interconnecting piping and the economizer inlet tubing must also be exposed to the same reducing environment. Observation of Table 3 indicates that no serious FAC failures have occurred in fossil plant mixed-metallurgy systems; however wall loss associated with FAC has been observed in fossil plants, and serious failures have occurred in nuclear plants with mixed-metallurgy systems; thus the carbon steel components in mixed-metallurgy feedwater systems must be subjected to the same rigorous FAC programs as for allferrous systems. Finally in this fossil/industrial plant example, Table 4 provides a comparison of the optimum feedwater chemistry for all-ferrous and mixed metallurgy systems. 4. Concluding remarks Although FAC remains a major problem in nuclear, fossil and industrial plants, solutions and approaches are available. The thorough understanding of the mechanism in feedwater systems which have all-ferrous metallurgy and mixed-metallurgy allows the proper application of inspection and cycle chemistry approaches. The recent studies involving the effects of oxygen and reducing agents have led to the use of ORP as a main chemistry controlling parameter. Several computer programs to predict FAC are available, and together with the optimum inspection techniques and cycle chemistry should provide the tools to implement
FAC programs which will ensure an adequate level of safety.
References [1] Chexal VK, Horowitz J, Dooley RB, et al. Flow-accelerated corrosion in power plants. EPRI TR-106611R1, July 1998. [2] Dooley RB, Mathews J, editors. Fifth International Cycle Chemistry Conference. EPRI TR-108469, October 1997. [3] Dooley RB, Mathews J. The current state of cycle chemistry for fossil plants. Fifth International Cycle Chemistry Conference. EPRI TR108469, October 1997. [4] CHECWORKS Computer Program User Guide. EPRI TR-103198P1, December 1997. [5] CHECUP, a CHECWORKS Application for FAC Evaluation of Fossil Power Plants User Guide. EPRI TR-107066, September 1996. [6] BRT-CICERO. Flow-accelerated Corrosion Software. EdF, December 1997. [7] WATHEC/DASY. Flow-accelerated Corrosion Software. Siemens/ KWU, 1991. [8] Chexal VK, Munson DP, Horowitz JS, Randall GA. Recommendations for an effective ¯ow-accelerated corrosion program. NSAC202L-R1, EPRI, November 1996. [9] Dooley RB, Tilley RM, Chexal VK, Horowitz JS, Munson DP. Guidelines for controlling ¯ow-accelerated corrosion in fossil plants. TR108859, EPRI, November 1997. [10] Chexal VK, Dooley RB, Munson DP, Tilley RM. Control of FAC in Fossil, Co-generation and industrial steam plants. International Water Conference 1997. Paper IWC-97-57. [11] Tilley RM, Dooley RB, Brett CR. Recent developments in managing FAC in feedwater pipework for fossil-®red power plants. Joint Power Generation Conference, 1998. [12] Filer S, Janick M. ORP provides versatile water treatment power engineering, vol. 102, #11, November 1998. p. 50±6.