RIV000129 Submitted: October 11, 2012
G. J. Bignold, BSc, PhD, K. Garbett, BSc, PhD, R. Garnsey, BSc, PhD, C.Chem, FRSC, and L S. Woolsey, BSc, PhD, CEGB, Leatherhead
Erosion-corrosion of carbon steels has been experienced in the steam generator and secondary water circuits of many reactor systems. Damage has occurred under both single and two-phase water flow conditions, and is associated with severe fluid turbulence at the metal surface,. In the most s.e~re cases, this can lead to very high metal wastage rates (>lmm/year) • and consequently rapid component failure. ·'Ihe available experience, previous research and current understanding of 1the phenomenf?D are reviewed, and both experimental and theoretical work in progress at CERL is / described. 'Ihe pH dependence of the phenomenon under single phase conditions at 1480C is r"ported., and by using hydrodynamically well characteri&ed specimens, the dependence of erosion-corrqsion rate on mass-transfer has been investigated. At 148°C, the rate has been found to vary as the eube of the ~ass transfer coefficient. 'Ihia· is in agreement with the ·predictions of a model of the process based on tbe electrochemical dissolution of magnetite. In order to make quantitative measurements · on the process, high precision bore •trology and surface activation of the test spaeimens has been used extensively, and these measurement techniques are also discussed. INTllODllCTION has occurred in the steam-water circuits of water ·1. · Nuelear steam generators have experienced a and sodium cooled reactors. As a result there is wide variety of corrosion related problems, and growing international interest in'erosio~ the vulnerability of individual designs to any corrosion phenomena (ref. 4-7). The present particular type of .corrosion damage can vary paper therefore attemptS to summarize current widely; In all cases, however, the econ~c experience and understanding of the problem, and .--..penalties resulting from such damage are consider describe erosion-corrosion work in progress at able, and there is therefore a st-coni incentive CEIIL. to eliminate such problems as far as possible. To this end, a wide variety of research programmes EROSION-CORROSION are in progress throughout the world. 4. The term erosion-corrosion is slightly -2. In many nuclear systems, corrosion bas misleading and the phenomenon is perhaps better resulted from the generation of aggressive described as flow assisted corrosion. As such solutions via solute concentration processes it is clearly distinguishable from pure erosion (ref.l). this is particularly true in the case or cavitation damage. of PWR steam generators, for example with the denting, phosphate thinning and tube sheet 5. Erosion-corrosion damase normally occurs at crevice stress-corrosion problems (ref.Z). locations where there is severe fluid turbulence adjacent to the metal surface, either as a 3. In the case of U.K. gas cooled reactor steam result of inherently high fluid velocities. or . generators, considerable effort has been directed the ·presence of. some feature (bend, orifice etc.) at understanding and eliminating the possibility generating high levels of.tutbulence locally. of corrosion damage resulting from solute Its occurrence is also uSually associated with the concentration under two phase flow and dryout use. of mild or carbon steal components. 'Ihe conditions, and the vulnerability of both Magnox attack occurs under both single and tvo-pbase and AGR steam generators to on-load corrosion water conditions, but not in dry steam, which is and stress corrosion has been reviewed very consistent with tbe general view that the process recently (ref.3). The need for stringent feed is essentially one of surface dissolution. It water chemical control was recognised and to is frequently, . iu though not invax:iably, ·date they have not proved to be·a problem. characterized by the occurrence of overlapping However, both Magno_x and AGR steam generators horse-shoe shaped pits. giving the surface a have been subject to an entirely different type scalloped appearance, as shown in Plate 1. of corrosion da-ge not dependent on any solute However. these pits are normally relatively concentration process, namely erosion-corrosion. shallow in comparison to the general metal Similar erosion-corrosion problems have also wastage in the area c011cemed. 'Ihe•oxide present been encountered in other gas cooled reactors on the corroding surface is normally very thin, .o- elsewhere, most notably in France and Japan, but 1 um or lass, and often eXhibits a polished ' the problems are not restricted to gas cooled appearance. Houever, heavy oxide deposition is reactor steam generators, and this type of damage sometimes oresent on adiacent_~~~~-Pi tube not s ...... PLATE 1. Erosion-corrosion damage produced under two phase conditions in a mild steel riser pipe from a Magnox steam generator. Flow from left to right.
l 1 l• f PLATE 2. Erosion-corrosion damage downstream of the orifice in a CERL mild steel orifice assembly specimen. Flow from left to.right.
PLATE 3. Metallographic cross section of specimen shown in Plate 2 in region of maximum . erosion corrosion loss. r 6 .J.----....
"--·
10 cOIIROSICIII-EROSION LOSS y u FEEDWATER pH ?.1,_ c .."'Ill . 9 ... =0 'i I 10 IDO CORIIOSIOM·ERDSION LOSS, mils/yew
FIGURE 3. Erosion-corrosion loss rate v pH for a rotating diae at 99°C (ref.26)
~-·- FIGURE 1. Temperature dependence of erosion-· corrosion losses under two phase conditions (ref.l9)
~'\.0• I'ATTIIII" Atre•ltta: W!LOCITY .. ~-I -~~ f-!- "--· <= ' AT eLaon I ...... ·rl· 'IILOCif'f'M f.;- t:Ln• f- __ j ·~ "Lanl-= t.-'hAL fLO• ~ JUGitaTJD• tUI'StaiAMOf -,-.r I JT.\G ..fiOJt ' 'DflfTS ~ ..Jt , •• Pt,.l .ltiiiiCTIGIII OIISUC:LI"I ·;·I . ..• •=il·-.---- fa DO\ t!:! :-~ ...... HCCNI:DnY IUP 1- FL... .,.., .. f&.OW ¥tLOCJTr $1'ACJIAtl011 -~ .Ill.,, .., f'Out1S ' 1- UXIWO ..IH ~ JOtltTI ... f l ~; FLOtt JlAGIIr.TIOII ... f'OINYIDUI ~~ ~~·:~~~"~·· •t.PYI.LOCtfY TO VOIITU .;;- f-- FOAMATtO• ----- ...... L1 •l'laatell1' 101 .... PLOW VILOCIYY .... n•QIATIO••• ·--· POIIITI Ju. ..no.'I'-ZOit'IAL 'tla.oon~no ~ ...... "' ...... VI&001'TCALCIIII.&'ICO COIIPUC&Tflt ,~ lllt ...... GUNDifA4...... fLOII' TNilOI,tQI TINIIetltiP&K'f r::::::'--rJUIO ...... , ~ OJ FQ c.,u.!~ ...... ·- ,.UIQ"......
FIGURE 2. Influence of flow path configuration FIGURE S. Arrangement of orifice assembly on erosion-corrosion damage under two-phase s.pecimens in autoclave flow channels. conditions (ref.lJ) ·
. 7 I _;....--.-~·· s~ffering erosion-corrosion attack, particularly Power Station, erosion-corrosion damage at the under two phase conditions. ·feedwater inlets downstream of the flow control orifices was compounded by flow·bypassing 6. under severe conditions, metal wastage rates through the gap between the threaded ends of the of 1 mm/year or even higher can be observed in restrictor tubes and the orifice carriers erosion-corrosion situations, so that component (re£.12). In some cases th1s fluid bypassing failure can be relatively rapid in the worst completely eroded away the restrictor tube end. cases. 11. In addition to problems within the steam Plant Experience generators themselves, erosion-corrosion damage ~ Under two phase conditions, erosion has frequently been encountered in wet steam corrosion damage within nuclear steam generators turbines (re£.13) and associated steam pipework has frequently occurred at tube bends, (ref.6), both the feedwater and steam-side of bifurcations or similar features in the steam feed heaters (ref.l4-17) and boiler feed pumps water circuit. Among the earliest reported (ref. 7). Clearly therefore the problems are instances of damage of this type were those at very widespread, and not restricted to any one the Tokai Mura plant in late 1968 (re£.8,9). type of nuclear plant. This station employs dual pressure drum re circulation type steam generators, and early Current Understanding of Erosion-Corrosion failures occurred at 214°C in·swan neck bends and Behaviour tube bifurcations at the outlet end of the mild 12. In spite of the widespread occurrence of steel L.P. evaporator tubes. Some failures also erosion-corrosion problems, as outlined in the occurred at tube bends in the subsequent riser preceeding section, relatively little experimental pipes to the L.P. steam drum external to the or theoretical work on the subject has been steam generator itself, and significant tube reported in the open literature. It is clear, thinning was reported for the last two return however, that erosion-corrosion behaviour depends bends of the serpentine evaporator tube banks on a number of physical and chemical variables. inside the units. Tube wastage rates as high as These are principally; materials' composition, 1. 3 mm/year were found in some cases. Up to the local hydrodynamic conditions including the time o£ the failures, the boilers had been effects of steam quality, temperature and water ·operated with hydrazine/ammonia dosing to give a chemistry. Any model of the process should boiler water pH in the rBnge 8.5 to 9.2. Some therefore be capable of explaining the detailed dosing. with Na3P04 was also employed to combat dependence of erosion-corrosion on these chloride ion (200-300 ppb) present in the water parameters. Their general influence on (ref.9). erosion-corrosion behaviour under boiler feed water conditions is summarised below. 8. Similar failures to these have occurred under steaming conditions in the mild steel 13. Materials' Composition. Erosion-corrosion economiser sections of British Magnox s tati"ons, damage is most frequently observed when carbon -.._. .. and in the evaporator sections of once-through or mild steel components are employed. Alloy steam generators such as those at St. Laurent I steels, particularly chrome alloy steels are and II. In the case of St. Laurent II, failures much less susceptible to. erosion-corrosion attack, occur~ed in the 180° return bends of the mild and austeni"tic stainless steels essentially steel serpentine evaporator to~ards the end of immune to damage. Relatively small amounts of the evaporation zone, at a temperature of about chromium in the steel improve its erosion 245°C (ref. 4, 10). As in the case of Tokai Mura, resistance quite markedly, although the degree the boiler feedwater was originally dosed with of improvement appears to depend on the severity ammonia and hydrazine to about pH 9.0. However, of the conditions. Thus in tests at 12ooc, more recently morpholine dosing has been employed involving impingement of a water jet on the because of its lower partition coefficient between sample surface at 58 ms-1, 2% Cr steel was foUnd water ·and steam and higher basicity at high to be at least an order of magnitude more temperature, which should maintain a higher resistant to damage than carbon steel, with solution pH at temperature (ref.ll). higher chrome steels even more resistant (ref.l8). However, practical experience with wet steam 9. Erosion-corrosion problems under two phase turbines and their associated pipework suggests conditions have also been reported to have 2!% Cr steel to be about four times more occurred in the steam generator units at Marcoule resistant to attack than mild steel, whilst 12% and Chinon 2 (ref.4, 10). Cr steel has proved to be virtually unaffected (ref .13). 10. Erosion-corrosion damage in nuclear steam generators under single phase (water) conditions 14. It is likely that other minor alloying or has commonly been associated with boiler feed trace elements such as copper, nickel, water tube inlets, and in particular those where manganese and silicon would influence resistance orifices have been installed to control the to erosion-corrosion as such elements are known boiler feed flow. Damage of this type has been to affect corrosion resistance of carbon and low experienced at St. Laurent II, with an inlet alloy steels to a wide range of aqueous feedwater temperature of about 12soc (ref. 4,10), environments. However, there appears to be no and at somewhat higher temperature (up to 2460C) systematic studies reported in the open in the case of the Phenix steam generators 1i terature. (ref. 5, 10). In the case of Hinkley Point 'B'
8
/ :;,.·-······-, ...· 15. Temperature. Erosion-corrosion damage is pH was less than 9.0, but attack was not most prevalent in the temperature range 50° to normally observed with pH >9.2. Similarly, the 25ooc. Fig. 1 shows the effect of temperature occurrence of erosion-corrosion damage in wet on relative erosion rates based on data derived steam turbines has been reported to occur only from damage occurring under two phase conditions when the condensate pH is below about pH 9.4 in Yet steam turbines (re£.19). This indicates (ref.l3,19). maximum damage to occur at around 180°C. How· ever, more recently it has been proposed that 21. The effect of pH on erosion-corrosion under single phase conditions, the maximum is rates has been studied experimentally by close to l40°C (ref.20). Limited studies on the Apblett (re£.26) using a rotating·carbon steel effects of temperature under single phase disc over the pH range 8.0 to 9.5 at 99°C in conditions have also been reported by Decker, deaerated water. The results are shown in Wagner and Marsh (ref,21) which would appear to Fig. 2, and indicate a tenfoid reduction in support this, but there remains some uncertainty wastage rate on increasing the pH from pH 8 to 9. in the precise variation of erosion-corrosion Similar reduetions in rate have also been rates with temperature. For example, rapid t:wo reported for jet impingement studies at 120°C phase erosion damage has frequently been observed (ref.27). at temperatures well in excess of 200°C (e.g. St. Laurent II), whereas the curve in Fig. 1 22. The effect of oxygen on erosion-corrosion would suggest the problem to be disappearing behaviour as ·suCh has not been studied in great rapidly at these temperatures. detail. However, iron release rates from carbon steel in neutral water at 1.85 ms-1 over the 16. Hydrodynamics. Erosion-corrosion damage temperature range 380 to 2040C have been shown has in general been observed at points of to decrease by up to t:wo orders of magnitude hydrodynamic disturbance in the fluid flow. with increasing oxygen content over .the range Under single phase conditions damage has <1 to 200 ppb (refs.23, 28-31}. It is to be frequently occurred at tube entries in preheaters, expected that erosion-corrosion will at least or downstream of orifices at boiler tube entries, qualitatively follow this type of behaviour. whereas un
20. In .studies of erosion-corrosion damage in is a variable factor accounting for the feed heaters (re£.14,15), it was found that effect of local geometry on the fluid flow. attack occurred predominantly when the feedwater Values of Kc in mm.s/m 10,000 hours· are
9 ( •'! ( I ..... 0
.- ..... , Main circutation pump I I \ T \ ;·-<'' r-o \ I ......
Tube Tube sample I sample2·
••it,__
p T
'r_ __ J;~
Venluri3
208 Kglh either sample, not both
P Pressure gauge T Thermometer TC Thermomeler/control F Flow indicator/conlrol · PC Pressure control L level indicator/control
FIGURE 4. Isothermal Rig flow diagram '-. '·..:
100
·; 80 ~ .. !00 u.. ~. ,; 0 ,. !E... 60 0 .. .. ";::"' •• ·~ .. )0 ~ =...X ..u .,... •• 3 20
IS
20 ,.
DEPTH •• m
.FIGURE 6. Activity/depth curve for 56 eo produced FIGURE 8. Variation of magnetite solubility in· an iron matrix by a 10.8 MeV proton beam with temperature and pH at 1 bar hydrogen inclined at 10° to the surface partial pressure. (Ref. 42)
7.SSO 1- r. ---·' 4.0 .. 1.450 ~ {1 ,,---~~ 7.750 1 I I .~ 3.0 f j E 7.150 1- i .!1 ... 7.45() ...... ~r· l ', r------!!:"' 1.! 1.15() pv r f.- I ! :::... 2.0 8 .., .. 7.150 i I ..."' :; 7.650 '!t: ~--·--..-.-·- ~ 3i "" ~ ...6 rv! 1.0 .. 1.15() ::: ll I ~ ! ! 7.156 - ! 1 •»" 7.150 L_V r--·~-7;_-..-
7.750 ~ r\ ; ·1. .i Iso •• TUBE DIAMETERS BEYOND ORIFICE 1,t50 ~ I J 't! f 1.950 .... ;,; FIGURE 7. Variation of mass ·transfer coefficient in a tube downstream of an orifice 110 160 tlt llO IDO ao DISUHCE FI\OH SPECittfM ounET."""
FIGURE 9. Erosion-corrosion loss profile in tube downstream of an orifice (Diametral circumferential locations shown. Orifice located 185 mm from specimen outlet) \ 11 ..
given in Fig. 2. behaviour under two-phase conditions.
c is the fluid velocity in ms-l Experimental Facility · 28. Experimental studies of erosion-corrosion K is a constant which the first term must are being carried out using a high velocity s exceed before erosion-corrosion is observed. isothermal water circulation loop, referred to A value of 1 mm/104 hours has been given by as the isothermal rig for short '{ref. 37). This Keller (ref.l9}. facility consists basically of a main circulation loop, a secondary water clean up loop and a Equation (1) does not include any influence from pressurizer loop, together with ancillary changes in water chemistry, although as indicated make-up/dosing and chemical sampling systems. A earlier, these have a very marked effect on the flow diagram for the rig is showa in Fig. 4. rate and occurrence of erosion-corrosion damage. It is also very doubtful that it can be applied in its present form to single phase erosion 29·. Four specimen flow channels are incorporated corrosion damage, since many instances of damage in the rig, two specimen autoclaves in the main have occurred under single phase conditions which loop, and two tube specimens within the secondary would not have been predicted by equation (1). polishing loop.
25. Di~cussions of some mechanistic aspects of 30. The rig is principally constructed of Type erosion-corrosion attack has been given by Romig 316 stainless steel, with the exception of the {ref.34) and Bohnsack (ref.35), who concluded that pressurizer vessel (2! Cr 1 Mo ferritic steel), the process is due to dissolution of the metal the heater elements (lnconel) and some parts of surface to give Fe2+ ions in solution, which are the main ci1:culating pump (Incoloy 825, stellite continually removed by the turbulent fluid flow. and ferobestos), The rig is design~d to operate However, both these authors restrict themselves over the following range of physical conditions: largely to discussion of the dissolution at 250C, which is well below the temperatures at Temperature, up to 350°C which el:o"ion-corrosion attack is normally encountered. At 25°C, Fe(Oa} 2 is normally Pressure, up to 21.78MNm-2(3160 lbf in-2) considered to be the co1:rosion product involved in the dissolution process in deoxygenated water, Autoclave up to 1031 kg h -l per autoclave but at temperatures higher than about 1oooc, this flowrates, is converted to Fe304 via the Schikorr reaction: Tube specimen up to 208kgh-l total (2) flowtate,
The rate of this reaction increases with tempera Bypass flowrate,up to 103 kg h-l ture, and magnetite is typically the phase observed on surfaces undergoing erosion-corrosion Pressurizer up to 20 kg h -l attack at temperatures above about 12ooc. As a flowrate, result, erosion-corrosion attack at these Once the rig '11ater has been -pressurized and '11ater higher temperatures has been attributed to rapid circulation achieved using the main pump, dissolution of the unstable Fe(OH) 2 intermediate (ref.36). · control of the physical operating parameters of the rig is largely automatic, with the variables 26. Very recently attempts to produce a model of interest {flow rate, temperature, pressure, of erosion-corrosion based on calculated mass water level etc.) being recorded by a dedicated CAMAC data logger. transfer rates and the solubility of magnetite have been made by GUlich et al. (ref.7). Wo1:k to produce a more satisfactory model is also in 31. The rig incorporates four methods for progress at CERL, and this is outlined in controlling the water chemistry, namely ion subs.equent sections. However, at present there exchange, chemical dosing, blowdown and deaeration. is no completely satisfactory model of erosion Data on the chemical composition of the water within the rig is derived mainly from continuous corrosion.~ehaviour which is capable of rationali sing the effect of all the diverse factors chemical monitoring of sample streams which can influencing the process. be.drawn from a large number of different sampling po1nts around the rig. The exception to this is CERL tROSION-CORROSION STUDIES the direct measurement of conductivity before and 27. The work currently in p1:ogress at CERL on after the.ion-exchange columns. To date, all the erosion-corrosion is directed at establishing a experimental work carried out on the rig has been consistent set of experimental data from which wit~ an ammonia dosed deoxygenated wate1: chemistry it is possible to make accurate-predictions of reg1me, and for these conditions it has been· found plant.behaviour, and to ·develop a satisfactory convenient to work with the cation exchange resins theoretical model of the process capable of of the mixed bed ion-exchange columns converted rationalizing the experimental work. At present, to their ammonium ion form. both experimental and theoretical studies are 32. In its present form, the rig is capable of concerned entirely with erosion-corrosion in operating within the following limits of physical single phase water, although it is to be hoped that the results of the work can be applied with and chemical con~rol parameters. certain limitations to erosion-corrosion
12 Temperature at test specimens ± 1°c hydr~dynamically. they can therefore be used for prft'cis~··correlation of erosion-corrosion and Flow to test specimens ± 1% mass transfer behaviour (see subsequent discussion). The particular ·specimens used pH of circulation water* ± 0.1 pH unit permit behaviour to be studied at five different potential erosion-corrosion sites; the -1 Conductivity of water after < 0.6 pS em tube inlet, the jet reattachment zone downstream cation exChange+ of the orifice, downstream of a tube expansion, and in two different diameter straight tube Dissolved iron in circulating < 10 pg kg-l sections (i.e. two different flow velocities). water at u.soc The specimens also have the _advantage that • being essentially straight tube test pieces, 1t -1 Dissolved active sili~a in < 10 pg kg is possible to use high accuracy bore diau~tral circulating wat~r at 148°C measurements to characterize the erosion loss profile throughout the specimen. Dissolved oxygen in circulating < 6 -1 water at 148°C )lg kg Erosion-Corrosion MOnitoring Methods 37. Simple weight change measurement;s are * Dependent on pH of circulating water, values possible on all the te~t specimens described, given for pH 9.0. At higher pH, the precision except the tube specimen Channels themselves. of pH control imp~oves, and dissolved Fe levels However • most of the effort to date has been fall. concentrated on monitoring damage produced in the orifice assembly spe~imens,· and this has + Upper limit of conductivity, ·due to very slow been done principally ~y the use of high sampling rate. accuracy bore diametral measurements, and thin layer surface activation methods. Test Specimens 33. A variety of erosion-corrosion test 38. Bore Metrology. Measurements of bore specimens can be incorporated into the isothermal diameter have been made on test specimens. using loop, using both the autoclave and tube specimen a "Diatest" internal bore lllQasuring instrument. flow channels. This instrument permits diametral measurements to be made with a precision of ±1 llmt and on a 34. The tube specimen channels are provided uniform tube surlace, the reproducibility was with couplings for the attachment of tubing . better then ±2 1Jm. The tubes used in the present between two points 2 m apart. Initially straight work are typically either drawn, or machined 3 1IDll bore mild steel tes.t specimens and stainless from b
13 .. .. • : ~ It .. .. ::! ... ;,. : II I I .. .. "' ~ I I I .~ j I -~ .. •L.~·~·~·~·~·~·~·~·-·_·-L--~~--~---L--~--~ lA u ... . ~ - - ftll-.• - - - - FICURE.lO. Velocity dependence of maximum FIGURE 12. Time dependence of erosion-corrosion erosion-corrosion rate observed downstream of loss for a surface activated specimen. an orifice at 157°C, and pH 9.05 :.-·'"
10.0 10 GEOMETRY- ORIFICE. z.n '""'DIA - TUIE • 8.:J.l •111.0. FlOW RATES- 9ZOit ~-I, • SLOPE - 570c a-•. o / -ZB1 ~b- 1 ." .. , ... 1: t •• .! it. I ~ .."' i! .f a ...i & .... t .··: f •••
a
a•~--~~~_.~~"~--~~~~-~·~ ..., L.-.L...-.L...-.L.-.L.-L--L--.L.-.L. 0.1 J.O 111.0 UUUtJUUUf.J MASS TRANSfER COEffiCIENT. • ,-I air ..
·FIGURE ll. Correlation of erosion corrosion FIGURE 13. gH dependence of erosion-corroSion rate downstream of orifice with corresponding rates at 157 C. Solid symbols, maKimum rates mass transfer coefficient throughout the from surface activated specimens, open symbols erosion-corrosion ~one average rates
14 . - . 56 . The bombardment of Fe with high energy protons where Shmax in equation (6) refers to the maximum produ~es 56co whick bas a half-life of 77.3 days, Sherwood number observed downstream of the and the principal y-ray emitted on decay has an orifice, and ReN the orifice Reynolds number. energy of 845 keV, The maximum depths of The overall variation of mass transfer activation for bombardment with protons at 10° coefficient K in the tube downstream of .the orifice is illustrated in Fig. 7. and 20° are around 35 and 70 )J1D respectively._ Deeper activation is possible by bombarding normally to the tube surface, or by increasing 46. The value of ~ in equation (3) may the incident proton energy. The total 56co clearly be determineM experimentally for any activity versus depth curve for bombardment at given erosion-corrosion situation, and for most situ~tions of practical interest will be very 100 to the tube s~rface is shOwn. in F~g. 6. low,_probably less than 10 ~g kg-1. However, 42. Loss of material from the specimen surface the concentration of iron in solution at the has been deterlDined in-situ by monitoring the oxide-solution interface cannot be so easily y-ray emissions from the sample using a evaluated. In !:he first instance, it might be scintillation detector placed in close proximity assumed that this term may be equated with the to the autoclave containing the active specimen. equilibrium solubility of the surface oxide which at temperatures above about 1oooc is ' These have perlDitted measurements of erosion loss to be made as a function of time with an accuracy usually taken to be magnetite. If however some metastable intermediate oxide suCh as Pe(OH) of ±0.1 ~m in the case of an activation depth of 2 is invoked, then a different solubility would 35 J,lm. be appropriate. 43. Full details of the experimental technique 47. The solubility of magnetite is known to be will be reported elsewhere (re£.39)~ dependent on temperature, pH and hydrogen Theoretical Work partial pressure (ref. 42). Fig. 8 shows the variation in magnetite solubility with pH and 44. If erosion-corrosion is controlled solely temperature at 1 bar partial pressure of :..-·- by the rate of mass transfer of Fe from the hydrogen (1585 ~g kg-1) derived from the data of eroding surface, then the erosion-corrosion rate Sweeton and Baes. However, these solubilities may be expected to vary according to are much higher than would be anticipated in operating plant, where the partial pressure of dm (3) hydrogen would be much lower (~1-5 pgkg-1). dt s K(Cs - ~) Under these circumstances the equilibrium Where K = mass transfer coefficient solubility of magnetite, when taken with the expected mass transfer coefficients is far too Cs = concentration of iron in solution at low to explain the observed erosion-corrosion the oxide-solution interface rates (ref.43). This analysis would indicate that the solubility of the surface oxide is ~ " concentration of iron in the bulk solution much higher than that expected for magnetite in· equilibrium with the bulk partial pressure of dm dt rate of metal loss. hydrogen. It is possible. however, that the solubility may be sufficiently enhanced locally 45. The value of the mass transfer coefficient by the high equivalent partial pressure of K varies with the local hydrodynamic conditions. hydrogen which results from the high local l~s de~ndenee on these is usually expressed in corrosion rate. Once established, the bigb d1mens1onless form using the corresponding local.solubil~ty in turn assists in maintaining S~erwood number Sh, where Sh = YJJ/D, D "' duct ~e h1~ eros1on rate. Electrochemically this ';hameter and D = diffusion coefficient for iron 1s equ1valent to the dissolution process · 1n solution. This is normally expressed in terms oc~urring.at r~latiyely negative potentials, of the Reynolds (Re} and Schmidt (Sc) nl.DIDers in wh1ch is 1n agreement with the general observation ·empirical correlations of the form that actively eroding areas are normally covered with magnetite, whereas nearby non-eroding Sb "' a~te 8 scY (4) surfaces are frequently eovered vith haematite. this possibility may be analysed theoretically where ca, Sandy are constants determined by in the following manner; e~eriment; Y typically has a value around 1/3, wh1lst the value of S is usually in t:be range 48. At equilibrium1 the dissolution of 2/3 to 7/8. Correlations of this type are magnetite to form Fe:t+ ions in solution (the a!ready available for a nl.DIDer of hydrodyn~c domin.ant specie~ under the conditions of interest) S1tuations of concern in erosion-corrosion and may be expressed as: two of particular interest: in the present. ~ork. are those for turbulent now in straight pipes Fe304 +2&20 +2H+ +2e :t 3Fe(OH)2 :t 3Fe2+ +60H (re£.40), and downstream of an ori~ice (ref.41). (7) These have the form: for which the appropriate Nerns~-equation is Straigbt pipes: Sh =0.0165 Re0.86Sc0.33 (5) E E9 ln fFe(OB) ] 3 (8) ..._. =- -.!!2F '- 2 Doi."DStream of an orifice: (6) (H12
15 which gives 813 51. In the case of specimens undergoing very rapid erosion-corrosion wastage, the films are [H+'] 9 f-2F(E - E ) l (9) sufficiently thin to exhibit interference colours. '[re2+ ] = 'With lower erosion-corrosion rates, however, the K2 expl 3RT J eroding surface is black, as for non-eroding areas ·of the tube surface.
[Fe{OH)2] tH+f 52. Micropitting of the tube· surface to a where K depth of about 5 pm is evident in the erosion 2 zone shown in Plate 3, and this is associated [Fe2+] with accelerated attack of the pearlite grains The cathodic current ic of the corrosion reaction of the stee~. Effects of this type have also resulting from hydrogen discharge at the surface been observed ;in plant specimens. of the magnetite film may be expected to vary exponentially. with the surface potential E of the 53. Most of the work to date has involved the film, as follows: use of mild steel orifice assembly specimens, and Fig. 9 shows a typical erosion-corrosion loss profile downstream of the orifice, obtained ic - FB exp (10) ~ {-a; ) using ~e bore measuring technique outlined If the anodic current ia at this potential is previously. The general similarity to the mass limited by the rate of removal of Fe2+ ions transfer profile shown in Fig. 7 is immediately from the. surface. a!J·d since i + ic = 0, apparent. However, it is clear that the . 8 straight tube losses are quite small, whereas Fig. 7 shows the mass transfer coefficient decays asymtotically to that appropriate to the straight tube, which is about 1/3 to 1/4 of that at the FB exp (- 11 (11) i':) mass transfer maximum. It is important to note a = 1 then substituting however, that the maxima in. botb curves occurs eliminating E gives approximately 2 tube diameters beyond the orifice. 54. Experiments exposing several specimens at 2 different flow rates under the same conditions 4K [H+] S may be used to establish the ·nov and hence mass (12) transfer dependence of the erosion rate, and K 3B2 Fig 10 shows the velocity dependence obtained at 2 1480c using pairs of specimens at three different flow rates. The slope the plot indicates a. v2 2 In this case the Fe + solubility of masnetite at dependence of -erosion rate on flow, which the surface is dependent on the square of the mass according to equation (6) would .indicate a transfer coefficient K, giving an overall dependence on mass transfer coefficient cubed. dependence of the erosion rate on the cube of Further confirmation of this K3 dependence is the mass trans.fer coefficient, through equation shown in Fig. 11, where the erosion loss (3). profiles of individual &Pecimqns have been compared point by point with the corresponding mass 49. This treatment may be extended to inClude transfer profile of the type shown in Fig. 7. all solUble iron species under tbe conditions From this it is seen that not only do the of interest, and the effects of a non negligible maximum losses downs tre- of the orifice· conform bulk concentration of iron. 'Dle expressions with the K3 dependence, but the erosien-corrosion become 1110re complex in this case, but still rates over nearly the Whole profile of the indicate a dependence of erosion-corrosion rate specimens correlate with KJ. · on the cube of the mass transfer coefficient (plus smaller terms in K2 and IQ ~ Further SS. lJhilst this alone does not substantiate analysis of the mechanistic aspects of erosion the theoretical treatment outlined in the corrosion is still under consideration, but this previous section, it does provide st:roag s~ort rather unexpected dependence of the rate on the for the type of mechanism invoked, and indicates cube of the mass transfer coefficient is bo:m. that: further development of the theory along out by experiment. these lines shoUld prove very fruitful.
Results 56. Fig. 12 shows the erosion-corrosion loss of an orifice assembly specimen clovnstteam of. SO. Plate 2 shows the erosion-corrosion zone the orifice as a function of time, detenai.ned downstream of the orifice generated in a mild fr0111 the activity loss of a surface activated steel orifice assembly tes&: speci111E!1l. Although spot in the erosion-corrosion zone. It is the suriace loss at the erosion maxiJIIUIII is evident that Ullder the particular conditions relatively large ('~>lSO ).1111) • scalloping of the used, there i5 a substant:ial initia;:ion time sm:face, of the cype shown in Plate l bas not befcn;e any erosion-corrosion loss is obnrvecJ. yet developed. Bovever, the. oxide film present Cbce initiated, the erosion-corrosion· rate iD the eroded area is extremely thin, as shown rose rapidly to a high value, and then remained in Plate 3. constant for most of the remainder of the test (the reduction in rate towards the end of the
16 test shown in Fig. 12 is thought to be due to 60. Since mass-transfer coefficients can be changes in experimental conditions). This type calculated for a wide variety of hydrodynamic of behaviour has been observed on a number of situations, at least under single phase occasions. although the initiation time can vary conditions. it should. be possible to use t~idely with the experimental conditions, generally correlations of this type to predict plant being much shorter under more aggressive behaviour over a wide range of conditions. ·2rosion-corrosiou conditions. The cause of such initiation periods is not certain at present. 61. Increasing pH has been shown to markedly In some cases this most likely represents the reduce erosion-corrosion rates over the range t'ime taken to remove a thin oxide film produced 9.05 to 9.65• in agreement with other studies of during start-up of the rig, when specimens are the effect at lower temperatures. In many plant exposed to low flow for a few hours. In other situations, therefore, this op~ion'should prove cases it is thought that thin air formed oxides effective in controlling erosion-corrosion produced during welding of the test specimens damage. It is likely to be especially useful were responsible. Hovever, in some cases, an when other options such as materiala change or initial loss of a few microns has been Observed, oxygen addition are not feasible. after which no loss has occu=ed for up to 200 hours, before true e"rosion-corrosion attack has 62. Details of the mechanism of erosion been initiated with a continuing linear loss as corrosion damage have still to be established, a function of time. 'Ihis would suggest that but the use of surface activation ill the preaent: initiation is more complex than simply removing work has proved to be extremely valuable for a pre-existing oxide film, and may indicate monitoring losses in-situ. Using this technique changes occur in initially formed films under it bas been possible to establish the linearity erosion-corrosion conditions. ~f erosion-corrosion loss as a function of time, . after some iuitiation period, and it: will. 57. The pH dependence of erosion-corrosion rates undoubtedly be useful in studying erosion has also been investigated using orifice assemblies corrosion behaviour under transient conditions. and 'Fig. 13 shows the results obtained at 1480C. In conjunc.tion with electrochemical techniques, The upper limit of the data is essentially therefore, it should prove very valuable in derived !rem the maximum linear races observed elucidating aspects of the corrosion mechanism. using surface activated speci~ns. The ~ates derived f:::om other spe.cimans are. average races, ACKNOi,ii..EDGEl'lENTS which are in general l~~er as a res~lt cf a significant but unknown initiation time. The 63. We wish to ~~ank J.H. Ashford, C.H. de Wballey, erosion-corrosion rates decrease by a factor of D. Libaert and R~ Sale fo~ their assistance with about: 7 over the pH range 9.05 to 9.65, which is the experimental work described, and M.i~.E. Coney equi valent to a variation •Ti th [H+ll· 4. This is for helpful discussions on aspects of ~s-transfer a somewhat higher dependence than that seen by behaviour in turbulent fl~~. Apblett (ref.26) at 99°C, where the erosion corrosion rate varies as [H+j 1.0. 64. This paper is published ~~th the permission of the Central Electricity Generating Board. SUNHARY REFERENCES 58. Ccrrosicn resulting from Salt ~onc-entration caused by evaporation continues t.c be a :najcr 1. G."-~'1SEY R. Boiler corrosion and the cause of steam generator damage, particularly requirement for feed- and boiler-water chemical in Pt•'Rs and has been the subject of intense control in nuclear 5 taam generators. BNES international research. Erosion-corrosion damage Conference on lolater Chemistry of !-fuclear Reactor has occurred in a wide variecy of nuclear steam Systems, BN~S, London, 1978, pp 1-10. generators, but unlike corrosion resulting from 2. GAre~SEY R. Nucl. Energy 18, 117, 1979. solute concentration, relatively little work on 3. GAR.'l"SEY R. Reducing the risk of ~ater-s team the problem has been reported in the open corros'ion damage in U.K. Gas cooled reactor literature. Tne available experience, previous steam generators. Paper presented· to ADERP resear~~ and current ~T:ders~andi~g ci the meeting on Water Chemistry and Corrosion in the pbenrinr.:~c.n have be.an r6·vi.ew~d~ and CERL resear.c.h Steam-water Loops of Nuclear Power Stations, --·~~ t.h-:: :i~bje~t su~;;iz-sd.. Seillac, }~rcn 1980. 4. CH..ANU.UEI .L.. G.enerat:eurs de vapeur chau fes 59. &u gaz ~azobvr.. ique dans les centra.les nucleaires s turl.y .:-ro"£ ia!1-c.orrcs. ios:. 1~:el!:!vic~'!' '.!~1dc ·r s :i ~gla £t8:i~;..,;ti;j.es. Paper presented t::, A.DEkP JL-:t=eting phase conditions c.O~fiP.T;:chlE r,;fth th1~se t.rhich. ern ~ate:r Ch.z::ziet:~");- ~~d Ccrrc3lcr;. in the S-ce~"'il occur in plant. By· using test s-peci.~n::; whi~h ~.ct~-: I.c.ops cf !!v.-c.lear r0~~-n2:!r Statio.u.s:- Seill.=.c. are well characterised hydrodynamically, it has }!arch 19 8CL been possible to accurat:ely correlate erosion 5. F.OB!~ H,G., gAJ.QN G~ and 'VJBER T. Les corrosion rates ~ith the corresponding ma5s generateurs de vp.peur patn:· rea~: teurs a ne•Jt!-ons ~ransfer race. Under the particular conditions :ral)ides. Paper presented to ADERl' mee·t:ing en useu (i4S6c, pn 9.05j. it nas been found that Water Chemistry and Corrosion in-the Steam-Water t!~~ ;-ai:.;: varie~ as the cube a.i D.'l.e mas~-transfer Loops of Nuclear Power Stations, Seillac, cc•.a.ffi.cl.;;:·:t. Tnis t.itle:-cpeccsd result is Harch 1960. C!-=?1'l:: 17 possibles. Paper presented to ADERP meeting p. 2, 1971 on Water Chemistry and Corrosion in the Steam 36. BORSIG F. Der Maschinenschaden, 41, 3, 1968. Water Loops of Nuclear Power Stations, Seillac, 37. ASHFORD J.H. BIGNOLD G.J. DE WHAiiEY C.H. March 1980. FINNIGAN D.J. GARBETT K MANN G.M.W. MCFAll F. 7. GULICH J.F. FLORJANCIC D. and MULLER E. and WOOLSEY I.S. CEGB Report RD/L/N 126/79. L;erosion-corrosion dans les pompes d'alimentation 38. CONLON T.W. WEAR 29, 69, 1974. . et d'extraction. Recherches et choix des 39. FINNIGAN D.F. GARBETT K and WOOLSEY I.S. materiaux. Paper presented to ADERP meeting on to be reported. . Water CheliJist:ry and Corrosion in the Stea:ur-Wat:er 40. BERGER F.P. and HAI1 R:. -F. F-L. Int J. Loops of Nuclear Power Stations, Seillac, March Heat Mass Transfer, 20, 1185,. 1977. 1980. 41. TAGG D. J. PATRicK M.A. and tlRAGG A.A. 8. Japan Atomic Pwer Company Limited, Third Trans. I. Chem.E. 57, 12, 1979. technical collaboration report on S.~u. tube 42. SWEETON F.H.lind BAES C.F. J.Chem. erosion and corrosion at the Tokai Nuclear Power · Thermodynamics, 2, 479, 1970. Station, J.A.P.c. Report,November 1969. 43. CONEY ~ Pl:ivate Communication. 9. Japan Atomic Power Company Lilllited, Fourth Technical Collaboration Meeting on S.R.U. repair work at the Tokai Nuclear Power Station, J.A.P.C. Report, September 1970. 10. GAR.AUD J. Revue Generale Nuclaire, No. 1, p. 29, 1978. 11. BERGE Ph. and SAINT PAUL P., Electricite de France Report HC PV D 389 MAT/T.42. 12. PASK D.A. and HALL R.w. Nuclear Energy, 18, 237, 1979. 13. ENGELKE W. in 'Two Phase Steam Flow in Turbines and Separators', ed M.J. MOORE and C-H. SIEVERDING, ~lcGraw-Hill, p. 291-315, 19.76. 14. PERGOLA A.C. and PHILLIPS A- Heat Engineer ing, p. 72, September-October 1965. 15. PHILLIPS H. Foster Wheeler Corporation (Nev Jersey), Report No. TR-36, 1966. I 16. KELP F. MITT V.G.B. 49, 424, 1969. 17. DOR F. Revue GeneraleThermique, 166, 705, 1975. - 18. wAGNER H.A. DECKER J.H. a~d HARSH J.C. Trans. AS}~ 69, 389, 1947. I 19. KELLERH. V.G.B., Kraftwerkstechnik, 54, 292, 1974. -- 20. KELLER H. J. !nternationales D'Etudes Des Cent:-ralas El 18 I i