The Effects of Erosion-Corrosion on Power Plant Piping
Spencer H. Bush
President of the Review and Synthesis Associates
Member of the National Academy of Engineering
and the American Society of Mechanical Engineers
59th General Meeting in 1990
Summary: The following article is a part of National Board Classic Series and it was published in the National Board BULLETIN. (10 printed pages)
The following material represents the author only and should not be assumed to be the opinion of The National Board of Boiler and Pressure Vessel Inspectors or the American Society of Mechanical Engineers administration, staff or membership, unless so acknowledged.
Major failures have occurred in piping due to single-phase or wet-steam erosion-corrosion, resulting in injuries or loss of life as well as extensive plant downtime. Both nuclear and fossil power plants as well as petrochemical plants are susceptible to erosion-corrosion. Causes, possible corrective actions, where to look in susceptible systems, optimum nondestructive examinations and analytic techniques to predict remaining life are covered.
Significant variables include temperature of water or steam, pH, oxygen content of fluid, quality of steam, flow velocity, quality of oxide layer on inner surface of the pipe, and chemical composition of the steel pipe. Typical susceptible systems are feedwater and let-down lines in water and elbows, tees, etc., in wet steam. Other systems operating under comparable conditions also are susceptible to attack.
Computer predictive models such as CHEC and CHECMATE developed by EPRI (Electric Power Research Institute) aid in selection of areas to be examined. Typically, zero beam ultrasonics (UT) is used, although radiography (RT) is an option. Another option is to use crawlers and TV cameras to examine the piping inner surface.
Factors mitigating erosion-corrosion which are under the control of the plant owner are the pH(>9.0 is best), oxygen content (50ppb), and pipe material (a 2.25Cr-1Mo steel is very resistant to single-phase erosion-corrosion while austenitics are resistant to wet steam).
A new subsection of Section XI of the ASME Code and an accompanying Code Case cover suggested regions of examination, examination procedures, and analytic techniques to predict safety margins in attacked piping.
The failure of a feedwater line at Surry-2 on December 9, 19861 , surprised both the utility industry and the United States Nuclear Regulatory Commission (USNRC). A steam valve closure increased the pressure about 20 percent, which was sufficient to fail a 90° elbow in the balance-of-plant (BOP) portion of the 18-inch feedwater line. A section of pipe about two feet by three feet blew out resulting in complete separation. The reaction forces moved the pipe about six feet. Four of eight contractor employees who were in the area subsequently died of severe burns.
Unfortunately, the replacement piping at Surry has already lost 20 percent of its wall after less than one and one-half years2 . Rumor has it that pickled piping was used. If so, the rate of attack may be twice that of a steam oxidized carbon steel.
While the failure occurred in the nonsafety BOP, activation of the fire protection system led to shorting of electrical circuits and seepage of carbon dioxide and other chemicals into the control room.
An examination of the failed section revealed thinning to below one-tenth of an inch. The cause was determined to be single-phase erosion-corrosion. Significant parameters affecting erosion-corrosion are chemical composition of the pressure boundary material, pH level, temperature and oxygen content of the coolant, and coolant flow linear velocity and turbulence.
Flow velocity is illustrated schematically. The effect of pH is given as a combination of pH and flow velocity. Dissolved oxygen is given, and temperature at various alloy levels for pH=9.05. Finally, as chromium and molybdenum increase, wear rate decreases. The ferritic and austenitic stainless steel essentially are immune.
Subsequently, extensive erosion-corrosion was reported at the Trojan plant in the secondary piping inside containment6 . A reexamination of previous failures and review of new inspection data gathered worldwide revealed that single-phase erosion-corrosion was more prevalent than had been recognized. In fact, it represents a generic problem. A General Accounting Office (GAO) report on erosion-corrosion confirms that several nuclear plants have suffered single-phase erosion-corrosion. A USNRC Information Newsletter presents an update.
Clearly, wall thinning due to single-phase erosion-corrosion is a time-dependent phenomenon. Under extreme conditions with plain carbon steel piping, severe attack may occur in a relatively short time. With more benign conditions and/or higher alloy levels in the pressure boundary material, the attack may be delayed many years or may not occur during the plant life. Systems particulary vulnerable are the main feedwater and steam generator let-down lines.
While this "white paper" concentrates on single-phase erosion-corrosion, both wet-steam erosion-corrosion, and droplet impingement are addressed. As noted, these are less significant failure mechanisms in safety-related systems.
EROSION-COROSION, ITS CAUSES AND PREDICTED OCCURRENCE
Single-phase erosion-corrosion is most likely to occur in minimum flow-recirculation lines, downstream of flow control valves (angle valves in particular) and in elbows in close proximity to other fittings. Instances of single-phase erosion-corrosion have been reported in other fittings such as at the small diameter end of a diffuser, etc. Erosion-corrosion is caused by a complicated interplay of a number of parameters. A large body of experimental work has identified several key variables that influence the rate of attack. These variables are listed below with an indication of how they impact the material loss behavior.
||Increase if Variable Is:
|Fluid pH level
|Fluid oxygen content
||Such as to create much more turbulence
|Component chromium content
|Component copper content
|Component molybdenum content
The complexity of these variables and their interrelation are such that a mathematical model which considers all of the variables is required to make erosion-corrosion predictions with any accuracy. This predictive capability helps avoid wholesale, random and nonproductive inspection efforts. The CHECTM (Chexal-Horowitz-Erosion-Corrosion) computer program10 was developed by EPRI to meet the industry need. The general formulation of the model used in CHEC is a series of factors which, when multiplied together, yield the predicted erosion-corrosion rate. Since some of the factors are interrelated, the model is not linear.
Since the interrelation between these parameters was not initially apparent, the formulation was developed empirically. A large data base was assembled from various laboratories, and an optimum model was obtained using an iterative procedure. This model followed all of the experimental trends, and correlated well with the bulk of the laboratory data. This model was further refined by comparing the predictions of the model with wall thickness inspection data obtained from several nuclear and fossil power plants, and with further laboratory research (particularly to take into account various geometrical mass transfer enhancement factors).
The correlation between CHEC predictions and the plant inspection data has been quite good. The ability of the code to predict single-phase erosion-corrosion rates within a 50 percent band, given accurate input data, has been demonstrated. This agreement is much better than other known erosion-corrosion correlations, and CHEC continually is being made more accurate by data and analyses feedback from nuclear and fossil plant users and by the results of continuing research.
Both CHEC and CHECMATE are discussed later in this report. Both require a complex interaction of variables, so the usual ASME Code appendix is not considered to be a feasible approach.
CURRENT MITIGATION OPTIONS
Depending upon the extent of wall thinning, utilities are applying several options to rectify the problem. These include:
- implementing a water chemistry change
- changing piping design/layout to improve flow geometries
- repairing or replacing with more resistant materials.
Water chemistry changes are attractive in that they offer a means of prolonging the life of existing piping. Two water chemistry variables, other than temperature, have been shown to have strong effects on the rate of erosion-corrosion, namely pH level and dissolved oxygen content.
In pressurized water reactors (PWRs), increasing the pH level can significantly reduce the rate of erosion-corrosion. In systems containing only ferrous alloys, pH control in the 9.3 to 9.6 range has been shown to yield acceptably low rates of attack from typical carbon, low alloy, and stainless steels in power systems. In plants with copper alloys in the feedwater heaters or condenser, a lower range (8.8-9.2) is required because copper attack has shown to increase markedly when the pH is above 9.2. The PWR secondary water chemistry guidelines do allow for operation above pH 9.2 if individual plant experience shows that copper transport does not increase significantly. The adoption of morpholine rather than ammonia as the pH control additive has also been used to reduce the rate of erosion-corrosion.
Boiling water reactor (BWR) water chemisty differs significantly from that in a PWR. First, chemical additives are not employed routinely. Second, significant oxygen levels exist in the condensate, feedwater and steam trains. A database exists illustrating the beneficial effect of maintaining the oxygen concentration in BWR feedwater and condensate above the minimum value given in industry guidelines. Operation near the 50 ppb oxygen upper limit of the indicated achievable range could reduce the probability of flow-assisted corrosion in single-phase regions.
Replacement of carbon steel piping components with low alloy steel materials has also been used successfully to mitigate erosion-corrosion. Current data suggest that an alloy containing one-half to one percent chromium would provide adequate resistance in single-phase systems. Two alloy steels which are available in a variety of sizes and for which there have been considerable power plant experience are 1.25 Cr-0.5 Mo (P11 grade) and 2.25 Cr-1Mo (P22 grade). As these low alloy steels have almost the same mechanical properties at the operating temperatures of interest, replacement piping of this material can be installed with the same geometry and unit weight as the original carbon steel components. Additionally, the thermal stresses and nozzle loadings are of little consequence due to the similarities in the coefficients of thermal expansion. As a result, the substitution of either of these grades is generally straightforward, and any design analysis should be minimal for the same configuration. One disadvantage is that both the P11 and P22 grades require special considerations for welding, especially preheat and postweld heat treatment. However, these considerations are well documented and represent standard practices in the industry.
Austenitic steels also have excellent resistance to erosion-corrosion. Low carbon grades are preferable because of better intergranular stress corrosion cracking (IGSCC) resistance. The candidate materials are 304L, 316L, and 347L. These materials are readily available and do not require preheat or postweld heat treatment. The disadvantages of austenitic stainless steels are that piping reanalysis is required due to a higher thermal coefficient (1.4 x carbon steel); the bimetallic welds need special attention; and susceptibility to chloride stress corrosion raises concern over the chloride contaminants in thermal insulation.
EROSION-CORROSION IN SAFETY CLASS PIPING
There have been cases of wet-steam and single-phase erosion-corrosion in 1987-1988. NRC Information Notice No. 87-36: Significant Unexpected Erosion of Feedwater Lines, dated August 4, 19876 , discusses thinning of Class 2 feedwater lines at Trojan. Significant parameters were:
|Nominal Wall Thickness:
|Minimal Wall Thickness per B31.7:
Trojan had less than optimum values for material, oxygen, pH and flow velocity and wasn't too good with regard to temperature. Several regions were found to be minimum wall, including 45°, 60°, and 90° elbows as well as straight sections of piping. Seventeen of 37 fittings and 125 of 366 feet of piping examined by UT were replaced because they were below code minimums.
A HISTORY OF ASME ACTIVITIES LEADING TO CURRENT CORRECTIVE ACTIONS
Shortly after the Surry-2 failure, a letter from Robert Bosnak of USNRC was sent to John Fernandes, a senior vice president of ASME and the chairman of the Council on Codes and Standards(CCS). This letter requested appropriate action by ASME. Recommendation 4 from the review of the Surry-2 failure states in part:
"The American Society of Mechanical Engineers (ASME) should consider the need for providing appropriate guidance to system designers on the subject of erosion and erosion-corrosion in its conventional pressure piping and nuclear piping Codes and Standards. Additionally, the Subcommittee on Nuclear Inservice Inspection (SC XI) of the ASME Boiler and Pressure Vessel Committee and the ASME groups active in plant aging and life extension matters should be made aware of the need to consider requiring pipe wall thickness measurements in their respective programs."
The first portion of the above recommendation deals with B31.1 and Section III and will not be addressed here. Only the inservice inspection aspects will be considered in detail; Section XI has the lead role within ASME on matters pertaining to plant life extension, and members of Section XI have been kept aware of the continuing status of the wall-thinning problem. A letter from John Fernandes requested action by the appropriate ASME groups by December 1987. Section XI responded to this request in a letter to him. The Section XI approach was presented to CCS in December, after which they instructed Section XI to proceed along the lines presented in the letter. Basically, the Section XI position was to leave the BOP as a utility responsibility and concentrate on safety-related systems. Nominally these are expected to be Classes 1 and 2 for BWRs and Classes 2 and 3 for PWRs.
A Working Group on Pipe Wall Thinning was established covering the areas of nondestructive examination (NDE), wall thinning and analytic evaluation techniques, and criteria for the selection of areas to be examined as well as the specifics on how much to examine. The intent was to take maximum advantage of the work of others, particularly EPRI, so that a code revision could be made in a relatively short time period. We expected to take advantage of the output of the EPRI CHEC computer program or similar approaches to optimize selection of regions considered to be most susceptible to single-phase erosion-corrosion. The bases for the CHEC selection were developed in the previous section of this paper and are discussed in greater detail in this paper.
With regard to NDE, the situation is not clear because at least three distinct forms of erosion-corrosion attack have been observed. These are:
- uniform wall thinning;
- highly localized wall thinning, often in widely separated regions (the area of such thinning may be 20 to 50 square inches), and
- severe and highly localized axial slot-attack analogous to a sharp slit generated by a cutting tool.
The differences in attack pose problems to NDE on where and how to look. If radiography is used, more complete coverage is possible but it poses problems to the workers in the vicinity. Ultrasonics, ususally a zero degree beam, should detect thinning, provided that the correct regions and specific locations are examined.
Once wall thinning is detected, a decision must be made on timing of replacement/repair, or continued operation. Fracture mechanics procedures have been developed for relatively thin-walled sections typical of most piping. Hopefully, these procedures will permit a less conservative approach to replacement than the one used to date of replacing pipe if it is below Code minimum wall.
The most difficult problem will be putting in Code language the locations to be examined, since they will vary from plant to plant. Basically, all ASME Codes are deterministic so it is relatively straightforward to define a weldment for NDE. Unfortunately, there are several problems that do not lend themselves to a deterministic approach such as:
- What specific components should be examined and how often,
- What NDE technique(s) should be used, and
- How large a grid pattern needs to be applied to components for NDE.
An expansion of item 3 indicates a dilemma we face. One could apply a four-inch grid, which has been done by several utilities. However, some use a UT transducer at the nodal points only(<1 percent of the area), others use a serpentine pattern along the sides of each grid (10-40 percent), while a few utilities examine the area within each grid. Another aspect is that a four-inch grid could miss slot-type attack; this means we will be plowing new ground in the Code in attempting to optimize examination without applying excessive requirements.
SIGNIFICANCE OF PARAMETERS AFFECTING SINGLE-PHASE
EROSION-CORROSION IN SECTION XI CONTEXT
The PWR primary piping usually is cast stainless steel or stainless clad low alloy steel operating above 300°C (600°F). Essentially zero attack at the pH and oxygen typical of primary systems was noted, as well as a major decrease in erosion-corrosion rate as one approaches 300°C. On these bases, one should be able to eliminate any examination for erosion-corrosion of the primary system. Class 3 systems usually are low temperature and pressure, often using austenitic stainless steel. The probability of attack by erosion-corrosion is minimal.
Class 2 systems, particularly the feedwater, definitely are susceptible; however, replacement with Pll or P22 materials (1 Cr-0.5/Mo or 2.25 Cr-l Mo) grossly reduces the rate of attack at low oxygen levels typical of these systems. If the pH can be raised to 9.5, the rate is reduced a factor of 100-1000 compared to pH -9.0. Obviously, carbon steel systems operating at about 200°C at velocities >18ft/sec with lower pH because of copper heaters are susceptible to erosion-corrosion. Any systems with such conditions are problems. Higher pH or higher alloy steels mitigate the problem.
Much of the safety grade piping in BWRs is austenitic stainless steel, with limited or zero attack. Furthermore, BWRs operating near the upper level of 50 ppb oxygen should be relatively immune to attack of carbon steel. The controlling factors where examination will probably be required are flow velocity >18 ft/sec, oxygen levels near 5 ppb, neutral pH, temperatures near 200°C and little or no alloy content in the piping. Systems markedly outside these values should be essentially free from erosion-corrosion.
One possibility is wet steam so the steam line may be attacked due to wet-steam erosion-corrosion or droplet impingement. This may need to be determined on a case-by-case basis.
Essentially, no safety-related systems suffer from wet-steam erosion-corrosion. Generally, these systems are in balance-of-plant, in particular, near the large steam turbine. A possible, but improbable system would be the steam line near the turbine; however, the contained water should be a low percentage.
Work by Kunze and Allgaier11 simulated operating conditions. Steam velocities were 186-202 m/sec, temperature about 200°C, steam quality -75 percent (25 percent water), and test times to 1200 hours. As with other conditions the E/C rate tends to decrease after several hundred hours. Five steels were tested ranging from a carbon and low alloy steel to a high chromium stainless steel. As in single-phase erosion-corrosion, the rate of attack decreased rapidly with increasing alloy. In fact, 0.10 percent chromium in the carbon steel reduced the rate to well below that of 0.5 percent Mo steel with 0.005 percent Cr. An EPRI report NP-394412 provides an excellent overview of erosion-corrosion in steam piping. The wet-steam case is significant in BOP but trivial in safety-related piping systems.
A special and common case of wet-steam attack is where water droplets at high velocities impinge on metal surfaces. Elbows are particularly susceptible. Again, this is a BOP problem, usually near the large steam turbine. The usual corrective action is to increase the alloy content in the elbows. In fact, some utilities replace attacked elbows with austenitic stainless steel elbows which eliminates the problem.
SELECTION OF SUSCEPTIBLE COMPONENTS FOR NDE
The decision concerning the rank-ordering of various classes of components such as elbows, tees, bends, etc., from most to least susceptible to erosion-corrosion is very complicated. It depends on the interaction of several variables with weighting factors applied to each of the variables. If one knows enough about a given system, it might be possible to use engineering judgment to select the rank-ordering; however, because of the complexity, computer codes are usually the choice. Siemens/KWU has developed the WATHEC computer code. EPRI has both CHEC and CHECMATE. While these are the ones cited most, others also exist.
The CHEC and CHECMATE codes are typical of the complexity of the selection process. The details and complexity of CHEC are apparent in a paper by B. Chexal, et al., presented at an International Atomic Energy Association (IAEA) Specialists Meeting in Vienna, September 12-14, 198913 . A large body of experimental work has identified several key variables that influence the rate of attack. These variables are listed below with erosion-corrosion an indication of how they impact the material loss behavior.
||Increase if Variable is:
|Fluid pH level
|Fluid oxygen content
||Such as to create more turbulence
|Component chromium content
|Component copper content
|Component molybdenum content
The complexities of these variables and their interrelations are such that an empirical model which considers all of the variables is required to make erosion-corrosion predictions with accuracy. This predictive capability helps avoid nonproductive inspection efforts.
The CHEC code was developed to help utilities respond to the Surry single-phase erosion-corrosion piping failure. CHEC was designed to allow a utility to determine the susceptibility of a plant to erosion-corrosion and to implement a limited inspection program to validate the results of the program. If the program and the inspections revealed a problem, the utility would expand the number of inspection locations sufficiently to ensure that all damaged components were identified and evaluated. On the other hand, if no problems were identified, the user could be certain that the plant was not in imminent danger. Thus, the utility would be spared the expense of making many unnecessary inspections to demonstrate that there was not an erosion-corrosion problem.
CHECMATE addresses a more pervasive but less urgent industry need. Failures caused by two-phase erosion-corrosion are relatively common in the industry. These failures usually occur in small, nonsafety-related lines but are a constant maintenance and operational problem. CHECMATE is designed to help utilities plan and implement a two-phase piping inspection program which will:
*identify the susceptible locations in plant before failures occur
*allow the plant operators and designers to implement sound design solutions
*maintain a data base of past inspection results and replaced components to further aid in the inspection planning process.
It is expected that with time, these codes would become primarily inspection planning and data management tools to assist utilities with their ongoing monitoring programs.
1. IE Information Notice No. 86-106: Feedwater Line Break, 12-16-86; also Supplement 1.
2. Reference 1, Supplement 3.
3. G. Cragnolino, C. Czaijkowski and W.J. Shack, NUREG/CR-5156, Review of Erosion-Corrosion in Single-Phase Flows, April 1988.
4. G.J. Bignold, et al., Erosion-Corrosion in Nuclear Steam Generator, Proceeding of the Second Meeting on Water Chemistry of Nuclear Reactors, British Nuclear Engineering Society, London, pp 5-18 (1980).
5. H.G. Heitmann and P. Schub, Initial Experience Gained with a High pH Value in the Secondary System of PWRs, Proceedings of the Third Meeting on Water Chemistry of Nuclear Reactors, British Nuclear Engineering Society, London, pp 243-252 (1983).
6. USNRC Information Notice No. 87-36: Significant Unexpected Erosion of Feedwater Lines.
7. Action Needed to Ensure That Utilities Monitor and Repair Pipe Damage, Report to the Honorable Edward J. Markey, House of Representatives, United General Accounting Office, GAO/RCED-88-73, March 1988.
8. USNRC Information Notice No. 88-17; Summary of Response to NRC Bulletin 87-01, Thinning of Pipe Walls in Nuclear Power Plants, 4-22-88.
9. J. Bridgman and R. Shankar, Erosion-Corrosion Data Handling for Reliable NDE, SMIRT Post-Conference Seminar No. 3, Monterey, CA, August 1989.
10. V.K. Chexal, E.B. Dietrich, J.S. Horowitz, G.A. Randall, V.C. Shevde, and J.A. Thomas. CHEC (Chexal-Horowitz-Erosion-Corrosion) Computer Program User's Manual, NSAC-112, Electrical Power Research Institute, Palo Alto, CA, Final Report, February 1988.
11. E. Kunze and W. Allgaier, Erosion-Corrosion in Wet Steam, VGB Kraftwerkstechnik, No. 1, January, 1985, pp 62-67.
12. G.A. Delp, J.D. Robison and M.T. Sedlack, Erosion-Corrosion in Nuclear Plant Steam Piping: Causes and Inspection Program Guidelines, EPRI NP-3944, April 1985.
13. B. Chexal, et. al., Technology Development by US Industry to Resolve Erosion-Corrosion, IAEA Specialists Meeting on Corrosion & Erosion Aspects in Pressure Boundary Components of LWRS, Vienna, Austria, September 12-14, 1988.
Editor's note: Some ASME Boiler and Pressure Vessel Code requirements may have changed because of advances in material technology and/or actual experience. The reader is cautioned to refer to the latest edition of the ASME Boiler and Pressure Vessel Code for current requirements.