ML082530370

Vermont Yankee July 2008 Evidentiary Hearing - Applicant Exhibit E4-19-VY, Effects of Chemistry on Corrosion-Erosion of Steels in Water and Wet Steam

ML082530370
Person / Time
Site:	Vermont Yankee File:NorthStar Vermont Yankee icon.png
Issue date:	08/12/2008
From:	- No Known Affiliation
To:	NRC/SECY/RAS
SECY RAS
References
06-849-03-LR, 50-271-LR, Entergy-Applicant-E4-19-VY, RAS M-334
Download: ML082530370 (6)

Text

2. Effects of chemistry on corrosion-erosion of steels in water and wet steam Ph. Berge, J. Ducreux, and P. Saint-Paul, EDF, Moret-sur-Loing It has often been observed, particularly in steam production plants that numerous cases of degrada-tion of steels occur when in contact with water or wet st'eam circulating at high velocity in feed or discharge pumps, water reheaters, etc. (ref. 1). When the phenomenon occurs without any mechani-cal wear of the metal or the oxide from the impact of solid particles (abrasion) or droplets 0

(erosion), it is called corrosion-erosion. The phenomenon usually occurs between 100 and 250 C, as has been confirmed by an empirical study of the thermal and hydraulic factors which govern it (ref.2).

Corrosion rates can reach 1 to 2 mm/year, for a carbon steel pipe where water treated with ammonia circulates at about pE 9, at 200*C, and at a velocity of 5 to 10 m/s. Usually the remedy is either a change in the steel quality, for example, the choice of a chromium steel, or the lowering of the flow velocity of the liquid or the wet steam. But in certain cases the only possible action is by chemical treatment of the water. In fact the part played by the water chemistry on the development and even perhaps on the occurrenceof the phenomenon has been noted, but without it being possible to make a quantitative estimation and without the mechanism of this action being clearly established.

In this study we propose to evaluate the part played by the factors solely connected to the chemis-try of water, with respect to tie kinetics of the corrosion-erosion phenomenon.

MECHANISM OF THE PHENOMENON layer.

Whereas oxidation of steels in dry steam follows a parabolic development, at least when an homoge- The dissolving rate, following the reaction (1),

neous oxide formed adheres to the metal, in the for a~given pH and hydrogen concentration, is case of oxidation occuring in circulating water, proportional to Cn eq - Cn (ref. 8) where Ce is it is quite different. This difference is due to the concentration of ferrous iron in equillbrium the fact that ferrous oxides which are formed (the equation being reversible) and C the concen-either directly by the action of water on steel, tration of ferrous iron at the surface of the or according to a more recent hypothesis by re- oxide layer, n depends on. the order of the reac-duction by hydrogen of the magnetite formed tion (1'). The following calculations have assu-(ref. 3-7), have a high solubility in water. One med, as a first hypothesis, n - 1, which leads can then observe a linear rate of oxidation after to a relatively simple expression of the corro-a relatively short time (Fig. 1) (ref. 4). The sion rate, by stating that the mass transfer slope of the linear curve depends on the nature (ferrous iron) towards solution is equal to of the steel, the temperature, the hydraulic con- X(C - CO) where X is the mass. transfer coeffi-Sditions and the water chemistry. This linear cient and C0 the concentration of ferrous iron function is due to the fact that the oxide layer in the solution.

formed attains a constant protection, i.e. a constant thickness in this case. The metal is dm A (C - Co) (2) thus oxidized in the form of soluble ferrous dt species. Dissolution by reduction of the magne-tite layer, by the equation dm E 2K (Ceq - C) (3)

Fe304 + (2-b)H+ + H2 from which one gets (1) dm 2Kk (C -C b 3 b) 2 O (4) dt". +2K+-X eq a

,is accompanied by the formation of a new magneti-te-layerlbf the same thickness by direct oxida- where K depends on the dissolving rate of the*

tijon.of the metal. This oxidation is itself magnetite under given chemical and hydraulic

.conditions.

accompanied by the formation of soluble iron which diffuses towards the exterior of the layer in a quantity more or less equal to half the mass The tests-in process should make it possible of oxidized iron. One can therefore assume that either to' verify. the proportionality between the corrosion is equal to twice the rate of dissolu- corrosion and the difference (Cea - Co), 'or to tion of-the external surface of the magnetite determine a factor.n giving a mofe complex axpres-sion of corrosion.

Water Cheinistir 1, BNES, 1980. Paper2 ..

- ".19

DOCKETED USNRC August 12, 2008 (11:00am)

OFFICE OF SECRETARY RULEMAKINGS AND ADJUDICATIONS STAFF U.S. NUCLEAR REGULATORY COMMISSI N Inthe Matterof *.A'*Jv V-, 4'e LLL-.-

Docket No.  ?-2 ' f ---- fcial Exhibit No. - - Vy OFFERED by @aricemee Intervenor-NRC Staff Other IDENTIED onikr*Z(O1 witness/Panel M LC Aon Taken: REJECTED WITHDRAWN en:[ý

,,egortar/Gerk

Chemistry and corrosion in steam generating circuits DISCUSSION conditions of a once-through steam generator of

- These hypotheses only take into account the an electric power plant, which developed severe mass transfer of soluble iron in the aqueous corrosion-erosion at the end of the evaporation phase in the vicinity of the surface. It can zone, the steam quality being about 90 % with an either be a diffusion in pressurized water, or ammoniated water at pH 8.9 - 9.

in the case of wet steam, a renewal of the liquid layer, necessarily present if this mechanism Table 1 shows the corrosion rate in different pH is involved. according to (4), for C0 = 0. The pH is raised either with NH3,. at different concentration, or

- It is not applicable to any mechanism involving with morpholine, the dissociation coefficient of a mc-hanical~degradation of the oxide. which at 250*C being higher, and the partition coefficient between water and steam more favou-

- In the case when the solution is saturated in rable than for NH3 .

ferrous iron, for example under isothermic and static conditions (autoclave), the long term The measurement of hydrogen produced in the gene-corrosion rate is practically nil (Fig. 1). rator has made it possible to ascertain that cor-relation with the calculation was reasonably good

- When the solution contains ccmplexing agents, over the whole range of pH tested around the for example chlorides, the value of C cannot be working pH level having shown corrosion (Fig. 2).

considered as a resultant of tha equilibrium of the reaction (1), and the corrosion is much EXPERIMENTAL STUDY higher. The Ciroco test loop (Fig. 3) makes it possible to evaluate various physical and chemical factors

- In the presence of oxygen, ferrous iron oxidi- governing, corrosion-erosion in pressurized water; zes into ferric iron which has a very-low solu- Most of the tests made to date have investigated bility and a different protective oxide forms the behaviour of mild steel (either preoxidized which can markedly slow down the corrosion rate or not) exposed to a flow of demineralized, (ref. 9-11). deaerated and alkalized water.

- If the metal is protectec by an oxide formed Before entering the test section, water is alka-under not very turbulent conditions, starting up lized with ammonia or morpholine at a pH within of rapid linear kinetics can take a certain time, the range of 8.5 to 9.7. It is also conditioned called the incubation period of the phenomenon. .with 20 ppb of hydrazine, pressurized at about 15 MPa and heated to 2250C. Flowing through a EVALUATION OF THE PART PLAYED BY pH ON CORROSION converging nozzle the jet of water speeds up to The value of Ceq depends on the pH of the solu- 60 m/s and impinges the flat surface of the test tion, the hydrogen concentration, and the tempe- specimen at an angle of incidence of 45*. Four rature similar nozzles are set up one after another-so that four samples can be tested at the same time C = KI (H) 2(H2)1/3 + K2 H+ (H2) 1/3 + under reasonably comparable conditions. Aftez K3~ K4 1H)

/3 cooling, the full water flow is polished in a (H2) /3 + (H 2 mixed bed ion exchanger, so that any complexing K3 agents (chloride, sulfate) and iron compounds (oxides, hydroxides, etc.) are removed as far as The value of the constants has been determined possible. It is then'necessary to adjust the pH by F.H. Sweeton and C.F. Baes (ref. 3).

with ammonia or morpholine as required.

As an example, we have calculated the values of The rig tubing and the main components are made Ceq in.boiler water treated with ammonia or mor-of stainless steel or Inconel 600.

pholine, at 250*C. This casecorresponds to TABLE I Ratio between corrosion at various pH under steam generator conditions of a graphite-gaz reactor (2500C - 90% steam) and in water at 2500C and that obtained at neutral pH (rf. 12.13).

pH at 250C 8.5 8.8 9.0 9.1 9.2 9.3 9.4, 9.5 9.6 With ammonia (steam quality 90%) 0.88 0.77 0.64 0.58 0.50 0.43 0.37 0.31 0.27 (steam quality 0%) 0.55 0.50 0.39 0.34 0.29 0.25 0.22 0.20 0.18 With morpholine (in water or wet steam ) 0.50 0.31 0.23 0.20 0.17 0.16 0.14 0.13 0.13 The partition coefficient of morpholine in water and steam being taken as 1.

9n

4 Total corrosion (mg/dm2) loin/a 10000 8lam SMI/s Soo ,,X . 2 s nozzle xyt.--. *test

• I

• in autoclave Time (h) 0 1000 1500 200O W

500 0

Figure 1 Effect of flow rate on corrosion in circulating 300 C water of low alloyed Mn Mo steel [4]

Heat exchanger

• Mixed bed: :Htxca Additives i~njection Circulatlng heater pum Measures of pH, conductivity, 07 H 2. N2 H4 , Fe, Cl0 S0 4 Figure 3 :Scheme of CIROCO LOOP and lost section.

H 22production img/h) 1,5 X Measured 0 Calculated Loss of metal S pH at 250C alcalizing 9.2 ammonia A, morpholine M agent:

400. _ preoxidized specimen not preoxidized

• specimen +

\./

0.5 0

pH at 250 C 6.0 6,5 7,0 100 20O Figure 2 ; Effect of pH on the H2 production in a steam generator Figure 4 : CIROCO LOOP corrosion curves.

of a graphite--gaz reactor. "

21

Chemistry and corrosion in steam generating circuits Under the above conditions, the loss.of metal obtained during a few hundred hours by corrosion-erosion is easily measurable by weighing each steel specimen. By continuous or recurring water sampling, it is possible to check the variations of the following chemical parameters : pH, catio-nic conductivity, chloride, sulfate, iron, hydra-zine, dissolved hydrogen and dissolved oxygen concentrations.

The steel surface preparation determines two dif-ferent types of corrosion versus time curves (Fig. 4) : when the steel has not been preoxidi-zed the loss of metal due to corrosion-erosion is proportional to the time ; in contrast when

-the steel has been preoxidized by exposure in high temperature static water, corrosion-erosion (a) " (b) starts after some delay.

Figure 6 :CIROCO LOOP specimen after testing Taking into account the above theory, one can a - Macrograph (after 80 h at PHol8.5 with ammonia) b

• SEM examination (selection atack of the peilite).

assume the rate of metal loss due to corrosion-erosion strongly depends on the pH. When the steel specimens have not been preoxidized, for Working conditions at the test section will be a given temperature, hydraulics conditions and as follows :

hydrogen concentration, when the water has not been contaminated by ccmplexing agents, the - Wet steam quality 80 to 100 %

corrosion-erosion rate depends linearly on the - Temperature 200 to 250 5 C ferrous iron concentration at equilibrium, gien by thepH (Fig. 5). - Flow rate 50 to 100 kg/h

- Impingement velocity on target 50 to 70 m/s Ferrous ions CO oncentrationat equillbrium (1) Scale,:A B C (ppb) pH at2250Cat 25°Cat 25PC (1) This concentration is calculated as a . 6.0 3 The water will be demineralized, deaerated and function of pH at 2250C (A scale) 6 conditioned before being turned into slightly 1 andCscales show the PH t2°C 2 0 2.0. necessary to obtain a given pH at superheated steam. Then the steam will be par-22 5°C (A scale) either with morphollne - ' tially condensed and the emulsion sent into the oe with ammonia 90 test section, then drained off after cooling.

An ultrasonic device will raa3: it possible to follow continuously the loss of metal.

1,0-

+ 6,5190 CONcLUSION The study undertaken has confirmed the very clk 7,0495 iQO; important part played by water chemistry on corrosion-erosion of steel.

12 The formula d 2K (C - Co) which makes 3 Corrosion I1mghI rate dt =K) ,AIC.q ef- o Figure 5

• Ci ROCO LOOP correlation between the corrosion rate and the it-possible to quantify the pH role on corrosion soluble iron concentration at equilibrium.

and which has been established taking into ac-count certain hypotheses on the kinetic order Some steel specimens which have undergone corro- of the Schikorr reaction, seems acceptable from sion-erosion either in the Ciroco test loop or the data obtained in the plant or in the labora-in power plants during normal operations look tory.

very similar when observed with a Scanning REFERENCES Electron Microscope ; in some cases the perlite

1. "Chimie de l'eau et corrosion dans les cir-seems to corrode more rapidly than the ferrite (Fig. 6). cuits eau-vapeur des centrales nucldaires".

Colloque ADERP-EdF, Seillac, Mars 1980.

Furthermore it has been observed that if there 2. KELLER H. VGB, Kraftwerkstechnik 54, 1974, 5,'

is chloride or oxygen present in the water, the 292-295.

corrosion rate is modified as indicated above. 3. SWEETON F.H. and BAES C.F. "The solubility of magnetite and hydrolysis of ferrous ions in REMARKS aqueous solutions elevated temperatures".

-In order to verify that the phenomena observed J. Chem'. Thermodynamics, 1970, 2, 479-500.

in the Ciroco test loop under monophasic condi- 4. BERGE Ph. M4canisme de l'oxydation des aciers tions are the same as those observed under dipha- dans l'eau a haute temperature et de la forma-sic conditions, a new program has been initiated. tion de ddp6ts d'oxyde. Congr6s d'Ermenonville, Another test loop in stainless steel giving wet 13-17 Mars 1972.

steam at the required temperature and steam:qua- 5. STYRIKOVICH M.A., MARTYNOVA OJ. et al lity, in under construction. Teploenergetika 19, 1972, H.9.S., 85-87.

6. STURLA P. S. Nationale Speisewasserkonferenz n-i in Prag vom 2 bis 4, Oktober 1973.

7. BIGNOLD G.J., GARNSEY R. and MANN G.M.W. High temperature aqueous corrosion of iron develop- m- 2KXT (cc-co) ment of theories of equilibrium solution phase transport through a porous oxide. Corrosion.

Science, 1972, Vol. 12, 325-332.

8. BERGE Ph., RIBON C. and SAINT-PAUL P.

"Effect of hydrogen on the corrosion of steels in high temperature water.Corrosion, 1976, 32, June, 223. n-2 mh

9. BOHNSACK G. Sonderheft VGB Speisewassertagung 1975, 15-19.

Asmtt

10. FREIER R.K. Wasser, Vol. 8, 1971, 443-458. Gradient at the origin Gradient -)

(Ceq -Cc) MKo)k

11. BRUSH E.G., PEARL W.L. Corrosion and corro-4KCo4AL ., '

sion product release in neutral feedwater.

i Corrosion,,Vol. 28, 1972, 129-136. co co+j,cq

12. POCOCK F.S. "Control of iron pickup on cycles utilizing carbon steel feedwater heaters". Proc.

Amer. Power Conference, 28, 1966, 758-771. n.13 m Asympltet

13. MESMER R.E. No publied results. Gradient-).

Gradient at thle origin (cea - Co) -

ttontehf ./

Addendum to Paper 2 ;aK~o~

XI 2n o' )

Mathematic study of diffusion and kinetic co Oaq equations for different values of n, the order of reaction. +

1/3 Fe 3 04 + (2-b) H* + .H2 Fe (OH)b Fig.l 4

.+ b) H2 0 b = 0,1,2,3 Considering the predomifiant ferrous species at equilibrium (for instance Fe (OH)3 in a basic medium according to the model of Sweaton and Baes) one can write down (1) I,= X (C- Co)

- 2K (Ceqn . Cn) or otherwise (II)th X (C Co)

Ceq = (Cn + (C-Co)) lln 2K Although the explicit function i = f (Ceq) can be calculated for n = 1,2,3, its study is not very easy forn - 2 and 3. One had better look for the shape of the curve Ai i f (Ceq) by study-ing the system (II) of parametric equations," f and Ceq being considered as functions'of the same parameter C. One has then to calculate the first and second derivatives du,"dCeq ; d 2 Ceq dc dc dc 1Z and deduce from them the concavity of the curve,-

the points of inflexion, and the behaviour at infinity.

The main results are summarized in the following diagrams (Fig.i).

The comparison between these theoretical results and-the trend of experimental measurements may help us to determine the order of n of reaction.

23