DEAERATOR CORROSION AN OVERVIEW OF DESIGN, OPERATION, EXPERIENCE, AND R & D COTAKAR JONAS, PLE Consultant Jonas, Inc. Wilmington, Delaware ‘ABSTRACT An industry-wide deaerator and storage tank weld corrosion cracking problen is currently ‘arong the most extensive and expensive probiens associated with fossil steam generation (ut! ity ‘and industrial). Over 302 of deaerators contain cracks which have been assessed. as corrosion Fatigue and/or stress corrosion. Causes of this problem have not been Found yet. Effects of material-Fabr cat ion-deslan~ Speration and "water chemistry need to be Invest !gated. The only correlation we have found so far ts that of the crack growth rate vs. tine in service: newer deaerators crack Faster than ‘the old ones, up to 15% of wall thickness/year. ‘An overview of deaerator destan, operation, ‘and corrosion problens and a brief description of RED activities are given. Possible material; Fabrication, design, operation and water chemistry effects end stress sources are Giscussed. Conclusions and recomendat fons are made which include’ the need for an expanded effort to determine the causes and develop Preventive measures for the alleviation of the Seserator cracking probien. IntRoDucTiON. Deaerators _(atrect heaters) are’ used. in. fossil and some nuclear power plantas Industrial steam plants, and. ship Propulsion. ihey remove dissolved gases” (nostly Sxygen and nitrogen) by Rechanical deseration and provide water storage capacity to give a Fesidence’ Eine at. “appropriate elevated Penperatures For” chemical oxygen. scavenging. In audition; deserater storage” capacity and hyorestatic need provide Flow “conditions for the bol ler feed pune. The operating temperature range 15103 to over 160°C (218 to ever 350°F) with the corresponding saturation pressure Up to over 0.96 Pa (ia0 pata). Another | type of oxygen renovel equignent 13 the deaerating heaters Tt reduces oxygen to about 0.0449/1 (40. ppb). The differences between Seoerators and deserating heaters are the degree of refinenent used In. water atomization, stean Serusbing action, and “venting "oF gases. Open feedwater heatepe, with a typical temperature ange (70 to 100°C) (160 to 210°F), are a1a0 used for dezeration. They os So aps contact deaersting remove oxygen only down to A typical run boiler between the location of @ deaerator in utility and once-through fossil cycles 1s Tow pressure and high pressure Feedster nesters before the boller feed pump (Ly 2)-. Steam which drives the deaeration is usually’ supplied From the IP turbine exhaust, the First LP extraction, or from the boller feed pune turbine exhaust (uring nigh load operat ion). and From auxilifary sources during startup, shutdown, ‘and not stenaoy. In “many” Industrial plant systems, the geserator 1s the only feedwater heater and Feedwater ts punped from the storage section of ‘the “deaerator directly through the econon'zer Into the Boller drun. Frequently, low pressure exnoust, low pressure flash tank, of blowsown Flash tank steam Is used for deagration. Also condensate (usually 40-60% of the Feecuater) Is, normally mixed with makeup at the deaerator In Systems with condensate return. There are many deaerator design configurations, horizontal and. vertical, with Gifferent Internals where the feedwater and steam Interact. Typical “designs encorporate spray, tray, and "combination spray/tray — heat ing sections. Figure | shows @ tio vessel horizontal tray deaerator used In most larger Industrial and utiTity systems. Host deaerator vessels are fabricated from carbon steel. Some, particularly marine stean propulsion deaerators, are made of copper alloys. Some deaerator vessels in industrial appl {cations are glass lined. Attempts to epoxy coat the Vessel 1D surfaces have not been successful. The coating peels off and there could be corrosion Under the coat ing. Figure 1 ~~ vtttity tray deserator corrosion crack locations um with typicalProcoedings of the Anerican Power Conference DEAERATION & OXYGEN SCAVENGING The mechanical deoerat ing action _usuol ly combines renove! of gases from the beter surfaces and. stean stripplng. Que to the gos volatilities. the “soses dissolved In water are raplaly trying forestebi Tah an equilibrium with the vapor-oms nisture In contact. Equtttoriun goa. concentration Tnwater' cy Ts according to. henty's laws cy * Sik where B 13 the partial pressure of gas a fg, hinry’s “constant (temperature depend) &. Since the gas concentration in water Is proportional to the partial pressure of the gas around the Viquid and the deaerating steam has very low contents of gases. gas concentration in water becomes very low, a Few ppb. Under typical Geaerator operat ing ‘conditions, 7 ppb _ of Gissolved oxygen Is achievable and Is guaranteed by the equipsent vencors. During deaerat ion, 90 ~ 984 of dissolved oxygen 1s easily renoved by mechanical separations the remaining 5-104 oF cxygen must be renoved by molecular diffusion. Heating -- Water in the deaerator must be heated to. full saturation temperature corresponding to steam pressure in the unit. Complete gas renoval Is not possible unless the liquid Is kept at botling temperature. Agitation -- Heated water must be mechanical ly agitated by spraying, cascading over trays, or atomization to expose maxima surface tothe Scrubbing stean, permitting complete release and renova! of gases. Thorough agitation overcones tendeney for surface tension and viscosity to retain gas bubbles and increases the rate of gas Giffusion fron the Ilquid to the surrounding steam. Intimate mixing of stean and water 1s essential ~ @ design requirement . == Enough steam pust pass through the water £0 false it to the boiling point and scrub Out and carry avay non-condensable gases after their release. Proper venting Is required. To reduce steam loss by vents, many deacrators are equipped with vent condensers to conserve eneroy and water. Both spray and tray deaerators should be able to renove oxygen to less than 7 ppb and the deseration efficiency should be good even at low Toads (see Figure 2). torre 0% i Figure 2 -- Expected dissolved oxygen at the Geaerator outlet vs. load for tray and sprey Seoerators. a Chemical _Oxwaen Scavenging -- which takes place in the deaerator storage tank depends on temperature, residence time, pH, and scavenogr conegntration. At temperatures’ above 200°C (40°F), hydrazine (which Is the most common oxygen" scavenger In utility appl feat ions) theemar ly decomposes (ZN, Hy Neti v2) and © portion of “it to, thereford, név “avalfable For Ehe react ion “with oxygen igieOs-92H0eNg) (spe Figuge 3). Most ceaeratorsCoferdee Getoe 200° aaah) i Figure 3--EFfect of _ temperature decomposition of hydrazine. Sourcs Wilcox. onthe Babcock & Dependence of restdence time, owen ratio’ and. pH has reported elsewhere (2, 4). Typical deaerator residence Eime Is 3 to'l0 minutes which fs sufficient for the oxygen scavenging “reaction to proceed. However, In some deslans, water from the deserat ing vessel may channel through the storage tank into ‘the outlet plee and” the actual residence tine may be very short and the scavenging poor. oxygen scavenging temperature, on the ine to Deaerators also remove other gases which enter the cycle with alr Inleakage and sone volatile” chemicals dissolved In. feedwater. Deserator “effectiveness in renoving carbon dioxide (a major” Impurity whieh can cause corrosion) {3 limited and depends on pi temperature, and deaerator and cycle design. At alkaline phy lover volatility carbonates, such as ‘amoniun "carbonate, form. These are not "deaerated” (see Figure 4 for percentages of C0, ‘gas removal vs pH). NHj 1s removed only in thé Form of NH,CH. “Like COR, It 1s. very soluble and requires 4 aiffusion “process. The removal of ‘anmonia occurs at more alkaline pl levels (see Figure 5). Ammonis, (Op, and other gases Frequently found in water®do not follow Henry’s Taw because when dissolved in water, they fonize. For exanple:Procendings of the American Pawer Conference ‘The chemical species present In the solution sepend strongly on pk. ‘As equations I+ 2, and 3 show, the solution of the gases in water is a reversible reaction. To obtain almost complete renoval of these gases, advantage may be taken of the common ton effect ‘That. Is, IF the concentration of on oF the tons on the right side of the equation Is Increased, ‘the reaction Is driven to the left, resulting In Bissolution of the gas. In the case of carbon Gioxide and hydrogen sulfide, hydrogen fon may be Increased by adding an ecid. Any bicartonate or carbonate in’ the water will be neutralized and the carbon dioxide formed can be renoved by eaerat ion. Ina Similar manner, armonia removal can be improved by Increasing hydroxy! fon concentration through the addition of caustic soda or phosphates. Properiy operated deaerators have beneficial effects on water and steam chenistry and cycle conponent corrosion control because they help to reduce oxygen in feedwater and, therefore, the pre-boi ler system and bol ler corrosion. It should be noted that the cycle corrosion depends on many other factors besides the oxygen concentration, such as pH, other impurities, matertals, and flow velocity’ ana turbulence (erosfon-corros ion) + Figure 4--Percent CO, available for removal (5). 28 pen eet EE ” Figure 5--Percent Ni, available for renoval (5). [CORROSION PROBLERS. An industry-wide deaerator vessel and storage tank weld corrosion cracking problem has been Giscovered after the 1983 deserator rupture Ino paper mill (6 1). Over 30% of utility ang ‘industrial deaerators (out of over tho ‘thousand inspected) have been found to exper tence this type of corrosion. In one ut! lity, 90% oF Inspected deaerators have cracks, in others, only 2 Few. There have been at least Five catastrophic Geaerator and other carbon steel pressure vessel follures. Similar problens have been experiences in West Germany over ten years ago (G, 9) and are just being discovered In the CEGB Utliity units In Engtand “end in KEMA units in Holland (26 inspected, 17 cracked). ‘The fracture characteristic are those of corrosion fatiaue in hot. water, In some cases of stress corrosion. Fitting and heavy oxide often accompany the cracking. Cracking Is more frequent below the water Tine and Inthe head to shell welds (see Figure 1).crack penetration through the wall was found In several deaerators. Causes of this problem nave not been discovered yet, but most likely there is a combination OF water chemistry, residual welding stress, and cycling service stress factors. Hvcreate cracking due to hydrogen generated by General corrosion and its diffusion into the higher strength heat affected zone is also 3 possibiity. MATERIALS, Host deaerators are fabricated from carbon steels (AST A 212 Gr. Cand ASTH A S15 Gr. 70) With sone stainless steel or cast iron (trays) internals. Although mich {9 known about carbon steel, there are significant gaps in. the knowledge related to deaerator cracking. There {s not a good explanation of the catastrophic failures’ where. the expected “leak before break" did not occur. This shows the need for knowledge of fracture properties and ductility of carbon steel weldnents after long ‘time elevated temperature exposure to hot water. Inpact energy, FATT, and fracture toughness data need to be developed to improve the knowledge of fast fracture. Fracture toughness may be es. low 28 50 KSI Vine Equal ly Important are corrosion properties 1n relevant environments, such as threshold stress eco threshold stress Intensity Kico, corrosion FaEFaue Vimits, AK, and crack propagation rates under corrosion Hitigue and stress corroston conditions. Also needed 1s tnfornation on the effects of pitting on the above properties. Knowledge of these properties would aliow the use of a more rations! approach in evaluation of NOT results and better prediction of service life and determination of inspection intervals, Carbon and low alloy steels are suscept ibleProceedings of the Anerican Power Conference to corresfon Inetuding: cracking In several environments, © Sodium hydroxide © carbonate ~ bicarbonate © carbonate - carbon dioxide © nitrate (© hydragen producing All these environments can exist in Seaerators under certain feedwater chenlstry and operating conditions. At high enough. yield strength (hard welds) or under high’ tensile stresses, even pure water and steam can cause stress corrosion and corrosion Fatigue cracking. WATER CHEMISTRY The water chemistry ang environmental factors which can lead to deaerator corrosion include high levels of oxygen and CO, low pH accumulation ef corrosion products 8m surfaces: and of ionic impurities within these oxides, and moisture and air or aerated water exposure during unprotected layup (comen U.S. practice) ‘One “design aspect “which can Influence Feedvater chemistry "Is the enormous length of sone deaerator storage tanks in large units, (up to 200 Ft.) with ‘the storage tank inlet ond outlet in the middie. When water purity” in the tank {3 allowed to deteriorate, such a3 Guring an unprotected layup (water saturated with eir)y the deaerator tank water can contain impurities For 9 long tine. Flow channel ing can occur In. two vessel units when a stream of water flows from the deaerator “vessel directly into the storage tank outlet and the impure water stays in the end sections of the storage tank. This can also Contribute to weld corrosion which is prevalent in the head to” shell welds. In some deaerator storage tanks, high pH deposits were found during failure analysis and sodium and silica were Ident fies by EDAX. Sets and caustic can concentrate on Generator storage tank surfaces by severs! mechanisms. When water Flashes into steam during even minor pressure drops, dissolved impurities Gistribute between the liquid and vapor according to their distribution coeFFicients (For molecular species) (ly 2). The distribution depends on the volume "ratio (or pressure). Wnen the metal surface temperature is higher than the saturation temperature (even by 3 Few degrees) dissolved impurities can concentrate from” any Yow concentration (ppm or ppb). to percent. Iron oxides which are allowed to accumiste can also concentrate corrosive impurities. STRESSES The Influence of design, resicval, and operating stresses and fabricat ion/welding Practices on the storage tank cracking 13. not known yet. Operating stress levels and the Frequency " of "the" Fatigue-cousing variable stresses and their origins need yet to be determines. During observations of deaerators during startup, it was found that vigorous bottling and vibrations” occur wien the Toad ant feedwater demand increase rapidly and. the deserator pressure drops. Violent boiling and storage tank vibration can also be observed during end after the shutdown depressur ization and particularly after a turbine trip. In one U.S. utility, @ horizontal deserator was. Found Jumped off" its foundation, which certainty Indicates significant forces. The basic deaerator design feature i.e. keeping large volune of water exactly. “st saturation, makes It very "sensitive to even small pressure changes. While pressure can change rapidly, tenperature and heat energy of the stored ‘water vary slowly. This can” and does result in a mismatch and rapid boiling “and condensat ton. One source of variable stress is the ‘water-piston" Instability, also known as the Kelvin-Helmoitz Instability (10). Under certain conditions of the storage tank vater” level and Froude number, repeated movenent of the ‘water-piston” ' occurs (see Figure 6). The "water-piston" 1s propelled along the tank and Impacts the tank head. As 't travels, it pushes water Into the steam transfer pipes and drains causing damage of the deaerator internals. Water-piston ang Its effects are more pronounces Jn longer deaerators, Resicual welding stresses are highest in the weld direction and can be as high as the weld yield strength. Complete stress rellef of o large Vessel is difficult and some residual tensiie Figure 6-—"Water piston" deaerator storage tank. Instability ino (CORRELATIONS. In the attenpt to correlate NOT data with a Few design and operat ional parameters, the only correlation we have found so far is that of the corrosion crack growth rate vs. time in service (see Figure 7). This correlation for the cracked Geaerator storage tanks shows that in the old cones, the crack growth rate ts typical ly about 1 wail thickness per year, while in the Geaerators Of the built 3 to 10 years ago, it is 6 to let Per year. There Is no correlation at this tine of Eracking vs pressure (see Figure 8) and cracking ys design to operating pressure ratio (see Figure 3). Most of the above data are for industrial deaerators and include results of the inspections performed by the Industrial NOT Co. anc results ‘of the Anerican Petroleum institute Survey.lt. reharh Figure 7—Deaerator storage tank weld cracking: crack propagation rate, Rf wal! thickness per year vs. time in service. Figure 8--Deserator storage tank weld cracking crack propagation rate, tof wall thickness per year vs. pressure. eae enon: a Figure S~Deserator storage tank weld cracking: rack propagation rate, 1-uall thickness per Year Wer‘design to operation pressure rato. There 1s also no significant difference between cracking of stress relieved and not stress relleved deaerators. The survey (LI) of a smell nurber of deacrators identifies possible correlation of cracking with reported Svater hanner" conditions. (OTHER PROBLENS Other deaerator corrasion problems include corrosion-erosion of vertical deeerators and general corrosion when oxygen concentration 1s high (up to 1000 ppb observed). Corrosion cracking. of carbon steel! hot water vessels is not confined to deaerator storage ‘tanks. Deaerator vessels, steam accumslators, P¥lt Sluun™ generators, CANDU Feedwater pipe. ond Cascade heaters experienced similar crackiag. ‘SAFE OPERATION AND. INSPECTION INTERVALS Safe operation can be assured vy NOT at proper Inspection intervals and repairs us needed. There are no standards or codes which Provide guidelines for interpretation of the NOT Gata. Sone are using the ASHE Pressure Vessel Code addition For general corrosion as a criteria for repair of cracks. This 1s not a very rational approach since local ized. corrosion cracking and Fracture are not governed by general corrosion. [ney b= Both ronservative or non-conservative, cepending on stress, environment, rate of crack Prosagation, and fracture properties of the mneterial. We have developed a calculation procedure which takes all the above effects Into account and considers both stress corrosion and corrosion Fatigue as crack propagation mechanisms. The approach 1s schematically shown in Figure 10 (iz). Use of this procecure may lead to Identification of root causes “oF deserator Figure 10—-Procedure for determination. Vife and NOT interval RESEARCH ACTIVITIES We don’t know about any coordinated industry-wide R&D project (worldwide) which 1s. addressing the various aspects of deaerator corrosion cracking and problems with distress of deaerater Internals. Major efforts by individual companies are almed at NOT and repair. We know of two companies, one EIT ity and one ol! company’, who have broader experimental prograns aimes at Tesolving the storage tank cracking problen. Host ‘major compantes and sone company groups (Aner ican Petroleum Institute) comp! led stat istics.Proceedings of the Anerican Power Conference The National Association of _ Corrosion Engineers (NACE) formed. @ Deserator Corrosion Task Group, T=7H-7 in the Spring of 1964. The objectives of the Task Group were to: o Coordinate invest 1gat fon end RED © Collect and clsseninate information (experience, correston,..+) 2 Provide guide! ines © Collect materials ~ design ~ operation ~ NOT = fatlure analysis cata for statistical analysis of causes of cracking ‘he Task Group started with Five Work Groups: WGa Non-Destructive Testing Was Operation & Hater Chemistry WGe Design, Materials, Fabrication, and Repair Wed Fallure Analysis WGe Failure Mechanisms & Corrosion Data Work Group “a" promptly publ Ished Proposed Guidelines on. NOT and Work Group "e" Proposed Guidelines on Deaerator Design, Haterials, Fabrication, and Repair (13). The NOT guidelines reflect experience with the wet fluorescent and nagnetic. particle — Inspection and surface preparation which are most widely used today. hey do not describe other techniques wich have also been successfully used, such as wet magnetic ‘inspection with white paint and black magnetic inky ultrasonics, eddy current, and acoustic emission. ‘Ihe Proposed Guidelines on Oeaerator Designs Noterials, Fabrication, and Repair give general Feconmencat ions including lowering of the Tesidual welding stresses (stress relief), and Timit of tenet ie strength of weld and base metal to" 70 KSI. According to this cocunent, cracks Violating minimum design wall thickness should be Tepaired. "(usually cracks 1/16 in. deep). A Fecomendation for the design shel! stress not to exceed SOR of the plate yield strength was Considered but it was not included in the ‘cocunent because supporting experimental dats ere needed. Tssuance of @ good design standard supported by material. and corrosion cata will not be an easy Job. It should be coordinated with the ASHE (Pressure Vessel Code), Heat Exchange Institute, and other organizations, none of whlch considers localized corrosion in’ thelr current practices Gaith the exception of corrasion fatigue considerations In the ASHE Pressure Vessel Code, Section XI, Reactor Vessels). Associated with the localized corrosion cracking considerations shduld be appropriate fracture criteria, knowledge of stress Fluctuations, local stresses, and characterization of the chemical environment. jhe leak-before-break failure mode of ductile naterials does not alvays occur In pressure Vessels for not’ water” and steam service. ‘The state of the problen and level of the collective knowledge were well summarized during the 1985 NACE. Symposiun (14). Additional papers were presented at the 1987 NACE Symposium a. CONCLUSIONS AND RECOMMENDATIONS Deaerators are beneficial for oxygen control and can be marginally effective In renoving (Op. Deaerators thenselves 1d to be profected against corrosion (general, stress corrosion, corrosion fatigue, and corrosion erosion) by feedwater chemistry and Impurity concentration control during operation ang_layup, and by control of operating ‘and design stresses. Deserator vessel and storage tank corrosion cracking ts among the most ‘extensive and expensive U. 5. corrosion problene associated with steam generation. nis type of cracking 18 not however, Vimited to deaerators and many types of carbon steel welded vessels and Piping for hot water and steam service are susceptible to It. The “fracture mechanisms appear to be corrosion fatigue end stress corrosion cracking. So Fars there Is limited knowledge on actual.’ stress levels and. stress Variations, composition of service and layup environments, and. fracture properties of the carton steel welanents ‘after short and long-time service. Residual welding stresses (as high as yield strength), vibratory stresses caused by “water piston", water hater. and routine pressure’ “changes, and corrosive impurities dissolved in Feednater or ‘concentrated on deaerator surfaces are all likely contributors to ‘the corrosion cracking problem. dn. industry-wide effort beyond the NACE volunteer activity is needed to resolve the multi-faceted problem of deserator cracking. Asa First step in the investigation of causes, design-operat ion-material data should be collected and correlated with crack growth Found by NOT. Corrosion fatigue and “stress corrosion properties need to. be comp! led and/or developed for appropriate sateralenvironment-stress conditions. racture properties such as K,- and J should be tested for new and! ged welSs and HAZ. Ultimately, localized —_corrosion considerations should be included In design and operating practices. NOT practices are adequate with the wet Fluorescent magnetic particle inspection being used most” often. All’ deserating section vessels. and Storage tanks shovid be periodically inspected or proof tested.REFERENCES 0. Jonas, “Transport of Chemicals In Steam Cycle", NACE Paper No. 245, Corrosion/65. 1985. AVP. Aschoff, Yo. Lees O.N. Sopocy, 0. Jonas, "Interim “Consensus Guidelines ot Fossti Plant Cvole chemistry", EPRI cS-4823 Project 2712-1, Final Report, June 1986. Wk. Pray, C.E. Schwelckert and B.M. Hinnich, "Solubility of Hydrogens Oxygen, Nitrogen, and Helium In Kater at Elevates Tenperatures". Industrial 8 Engineering Chemistry. 44, p. 1146, May 1952, Net. Dickinson, et al "An Experimental Investigation of Mydrazine ~ Oxygen Reaction Rates in Soller Feedvater", Proceedings of the” Aner ican Poxer Cont 19, 1957. Handbook’of Industrial Water Conditioning, Betz, 8th £d., 1980. TAPP] Journal, 66, No. 9. p. 1983. 83, Septenter Hatin, 41, Now 2, Pe Ty October 1983. 8. vn Is. F. J. Adansky. i. D. Teichnann, "Operating Experience with” Fuedwater Tanks". YG Kraftwarkstechlk 57, NO. [ly Nov. 1977, Ev tal, P. Hofstotter, "stress Situations in Feedvater Tanks with Deaerators at. the Top", VG —Kraftwerkstecnnik, 55, NO. 55, p. 321, 1375, R. Cranfield, Lem. Witkinson "Solving Fiuid-Flow Problems in Deaerators Using Scale Models", CEGE Research, No. 13, June 1962. 3.0. Robinson, “Deaerator Cracking Survey ~ Basic Design, Operating History, and water Chemistry "Survey", NACE Sympostum, Corrosion /86 March 19, 1986, Houston Texas. "Remaining Life and NOT Interval for Carbon Steel Deaerator Tanks", Jonas, Ince» Calculation Procedure. "Proposed. Guidelines Boiler Feedvater Heater Vessels and Storage Tank Welds", and "Proposed Guloelines on Deaerator Design. Materials, Fabrication, and Repair", Naterials Performance: 23) Novenber "1984. “Deaerator Cracking" NACE Symposium, Corrosion/86 Harch 19, 1986, Houston, Texas. "Cracking In Boller “Feedvater Eaulpment™ MACE Symposium, Corrosion/B7, March 10, 1987 Sen Franetaco. Dartne!l, and 0. H. For Inspection of