The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise

G Model CS-6637; No. of Pages 10

ARTICLE IN PRESS Corrosion Science xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Corrosion Science journal homepage: www.elsevier.com/locate/corsci

The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise D. Guzonas a,∗ , S. Penttilä b , W. Cook c , W. Zheng d , R. Novotny e , A. Sáez-Maderuelo f , J. Kaneda g a

Canadian Nuclear Laboratories, Chalk River Laboratories, Chalk River, ON K0J 1J0, Canada VTT Technical Research Centre of Finland Ltd., Nuclear Safety, P.O. Box 1000, Espoo FIN-02044 VTT, Finland c University of New Brunswick, Department of Chemical Engineering, P.O. Box 4400, Fredericton, NB E3B 5A3, Canada, d CanmetMATERIALS (CMAT), 183 Longwood Rd. S., Hamilton, ON L8 P 0A1, Canada e JRC-IET, Westerduinveg 3, 1755 LE Petten, The Netherlands f Structural Material Division (CIEMAT), Avda. Complutense, 40, Madrid 28040 , Spain g Plant and Systems Engineering Section, Nuclear Plant Engineering Department Hitachi-GE Nuclear Energy, Ltd., United States b

a r t i c l e

i n f o

Article history: Received 7 July 2015 Received in revised form 29 January 2016 Accepted 31 January 2016 Available online xxx Keywords: Steel Weight loss Oxidation High temperature corrosion SEM

a b s t r a c t A major challenge in supercritical water-cooled reactor development is the lack of a consistent alloy database. An international interlaboratory comparison test was organized to study the reproducibility of weight change data obtained for identical alloys under similar conditions in different facilities. This paper presents the test procedures, conditions, results, and additional characterization data. More variation in weight change was observed than expected. The scatter was small within the same laboratory, but large between different laboratories. Much of this variation appears to be attributable to differences in test facilities. The data generally agree on the relative ranking of the corrosion resistance. Crown Copyright © 2016 Published by Elsevier Ltd. All rights reserved.

1. Introduction The use of supercritical water (SCW) as the coolant in a nuclear reactor is the logical evolution of the current generation of water-cooled reactors, which generate almost all of the electricity produced by nuclear power worldwide. The use of SCW as the coolant in nuclear reactors increases the efficiency over that of currently operating nuclear power plants, decreases capital and operational costs, and decreases electrical energy costs. Water is a familiar and relatively safe heat transfer medium, and many power utilities already operate both nuclear power plants and fossil-fired SCW power plants (FFSPs), and can easily see the technical synergies. Selection of materials for the fuel cladding and other in-core components for a supercritical water-cooled reactor (SCWR) is a

∗ Corresponding author. Fax: +1 6135848219. E-mail addresses: [email protected] (D. Guzonas), Sami.Penttila@vtt.fi (S. Penttilä), [email protected] (W. Cook), [email protected] (W. Zheng), [email protected] (R. Novotny), [email protected] (A. Sáez-Maderuelo), [email protected] (J. Kaneda).

key challenge in concept development. While zirconium alloys have a low neutron capture cross-section and remain the preferred fuel cladding alloy choice from the perspective of neutron economy, it has been long known that these alloys experience unacceptably high corrosion rates in SCW [1]. Although some Zr–Fe–Cr alloys showed promise at 500 ◦ C, it has been shown [2,3] that these alloys can also experience breakaway corrosion at that temperature. In addition, zirconium alloys typically have poor high-temperature mechanical properties. As a result, Zr alloys are not acceptable for use as a fuel-cladding material in an SCWR. As ferritic steels typically experience unacceptably high corrosion rates at the temperatures that will be present in an SCWR core [4,5], the prime candidate materials for in-core use in the various Generation IV SCWR concepts are austenitic stainless steels, with some consideration being given to nickel-based alloys. Austenitic stainless steels and nickel-based alloys were extensively evaluated in the various nuclear superheat programs carried out in the 1960s [6–8]; Ru and Staehle [9] have written an excellent overview of the US work. Indeed, many of the materials issues currently being studied in the various Generation IV SCWR research programs were identified and studied more than 50 years ago. There is also an extensive body of work on materials for use in FFSPs [10,11] and

http://dx.doi.org/10.1016/j.corsci.2016.01.034 0010-938X/Crown Copyright © 2016 Published by Elsevier Ltd. All rights reserved.

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model CS-6637; No. of Pages 10

ARTICLE IN PRESS D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

2

supercritical water oxidation (SCWO), although the water chemistry in the latter application is much more corrosive than in a nuclear or fossil-fired power plant. The Generation IV International Forum (GIF) SCWR Materials and Chemistry (M&C) Project Management Board (PMB) identified two major challenges that must be overcome to ensure the safe and reliable performance of an SCWR: 1. Insufficient data are available for any single alloy to unequivocally ensure its performance in an SCWR, especially for alloys to be used for in-core components. 2. Current understanding of SCW chemistry is inadequate to specify a chemistry control strategy, as the result of the large changes in physical and chemical properties of water through the critical point, coupled with the as yet poorly understood effects of water radiolysis.

(source organization in parentheses): Canada (CMAT)—Alloy 800H; EU (MTA)—08H18Ni10T stainless steel; Japan (Hitachi)—310 stainless steel. The experiments were coordinated according to alloy, temperature regime, and major test parameters. It was decided to limit the data reporting to weight change only, due to differences in surface analysis and testing capabilities in different laboratories; participants were free to perform, and to report the results of, additional measurements if desired. In the context of this interlaboratory comparison test, weight change is defined as the final coupon weight after testing minus the initial coupon weight prior to testing. This is the number typically reported in general corrosion studies performed in support of the various SCWR concepts. This paper presents the experimental procedure used for the tests, the results obtained, and discusses their significance. 2. Experimental

To address these challenges, the GIF SCWR M&C Project Plan is made up of two work packages, one on SCWR Materials and the other on Radiolysis and Water Chemistry [12]. The Project Plan noted that: “consideration should be given to sharing heats of materials to create a more consistent database. A selected number of heats of each alloy should be designated, and a plan developed for coordinating testing and characterization on these alloys by various participating organizations. It is proposed that round-robin testing and characterization of identical alloys under similar test conditions be carried out to assess the reproducibility of the results. . .” As a result, the GIF SCWR M&C PMB organized an interlaboratory comparison test (often called a round-robin test) to study the reproducibility of data obtained in different SCW test facilities on the general corrosion of un-irradiated candidate alloys in SCW. The participants in the interlaboratory comparison test, and the designations used in this paper, were, by signatory: Japan—Hitachi Research Laboratory (Hitachi); Canada—Canadian Nuclear Laboratories (CNL),1 CANMET Materials (CMAT), the University of New Brunswick (UNB); EU Joint Research Centre—Institute for Energy and Transport (JRC-IET), VTT Technical Research Centre of Finland (VTT), Centro de Investigaciones Energéticas, Medioambientales y Tecnológicas (CIEMAT),2 MTA Centre for Energy Research (MTA).3 As this was a collaborative effort within the Materials and Chemistry Project Arrangement, it was agreed by the participants that the tests would not start until the Project Arrangement was signed by all participants. The Materials and Chemistry Project Arrangement became effective in 2010 December, and the final planning of the tests was carried out during 2011, with actions to finalize the coupon dimensions, nominate the alloys and identify the amount required for the tests, and identify the participating facilities. The final test conditions were agreed to in 2012 January. It was agreed that each participating signatory would provide coupons machined to the appropriate dimensions to the other participants, who would then follow a standard procedure to prepare (e.g., polishing) and characterize the coupons prior to testing, carry out the tests under the agreed upon test conditions, and then report the results. At the time of the signing of the Project Arrangement, the three signatories agreed to provide the following coupons

1

Formerly Atomic Energy of Canada Limited. CIEMAT joined the test program after it had started and were only able to test one of the alloys. 3 MTA later dropped out of the interlaboratory comparison tests. 2

The compositions of the three alloys tested in the interlaboratory comparison tests are listed in Table 1. Note that only a limited number of 310 SS coupons were available, and hence not all participants were able to test this material. The alloy 08H18Ni10T is equivalent to AISI 321 and DIN 1.4541. 2.1. Coupon preparation procedure A coupon preparation procedure was developed and agreed to by the participants. The work flow and its division between coupon supplier and test participant are shown schematically in Fig. 1. The various steps in the work flow are described in detail below. 2.1.1. Alloy identification The test alloys were to be identified by their commercial grade names. The name of supplier of test alloys as well as the chemical composition of the alloy was to be provided. All coupons were to be marked with a unique identifier. The coupon numbering scheme agreed to during the planning of the round-robin tests is summarized in Table 2. 2.1.2. Machining The test coupons were to be machined into flat blocks with the dimensions 10 mm by 20 mm by 2 mm. A machining tolerance of 25 ␮m (1 mil) was considered acceptable. A hole measuring 5.5 mm in diameter was to be drilled at one end of the coupons for mounting. If necessary the hole size could be modified to facilitate mounting of the coupon in the test rig. 2.1.3. Sample size measurement The dimensions of the samples were to be measured using a micrometer or a travelling microscope to an accuracy of 1 ␮m (10−3 mm) or better. 2.1.4. Polishing of the machined samples Each test coupon was to be ground using 600 grit sand paper first, followed by 800 grit paper, and then with 1200 grit paper. A surface layer of at least 30 ␮m in thickness, as determined by measurement of the coupon dimensions before and after polishing, was to be removed from each side of the coupon to reduce the possible effects on corrosion of machining-induced plastic deformation at the metal surface. 2.1.5. Degreasing/cleaning After surface finishing, the coupons were to be cleaned ultrasonically using acetone.

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model

ARTICLE IN PRESS

CS-6637; No. of Pages 10

D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

3

Table 1 Compositions of the alloys used in the round-robin tests. Alloy

Fe

Cr

Ni

Mo

Mn

Si

Al

C

Ti

Others

800H 08H18Ni10T SUS310S

bal bal bal

20.6 18.4 24.8

30.7 9.1 20.8

– 0.37 0.29

0.6 – <0.01

0.3 0.75 0.49

0.5 – –

0.03 – 0.048

– 0.61 –

– – S—0.001 P—0.22

Fig. 1. The work flow for the interlaboratory comparison test and its division between coupon supplier and participant.

2.1.6. Storage of test coupons If the samples could not be installed into a test autoclave immediately after cleaning, they were to be placed inside a N2 or Ar-filled container for storage. Alternatively, they could be stored in isopropanol to avoid formation of an oxide in air. The corrosion tests should be started within 3 days of polishing to minimize the formation of a surface oxide film. 2.1.7. Specimen installation The surface area of the specimen in contact with the specimen rack was to be minimized (line contact between specimen hole and the rack was recommended) in order to prevent crevice corrosion. The specimen rack was to be sufficiently insulating to prevent galvanic corrosion. The configuration of the specimens and specimen rack was to be reported. 2.1.8. Weighing of test coupons Before the coupons were placed into the test facility (e.g., loop, autoclave, capsule) and after the completion of the test, the weight of each test coupon was to be determined using a balance to an accuracy of 0.1 mg (10−4 g) or better. It was recommended that weighing of a specimen be repeated five times (at a minimum, three times). Datum of the weight was the average of these measured values.

Fig. 2 shows photographs of the finished coupons (Hitachi) prior to testing in SCW.

2.2. Test facilities and conditions The test parameters were developed by the participants and took into account as much as possible the limitations of the various test facilities (e.g., maximum operating temperature). The water chemistry (low conductivity water, no pH additive) was chosen based on typical testing practices in the SCWR community. The dissolved oxygen concentration was chosen as 8 ppm to simulate the production of oxidants by in-core water radiolysis [13,14]. The final test parameters selected are listed in Table 3. The test facilities ranged from static autoclaves to flow loops, and the main parameters of the various facilities are listed in Table 4. Further details on each facility can be found in the indicated references. In each test facility, SCW exposures of the different alloy types were carried out individually (only one alloy type in the test facility at any one time; CNL, CMAT) or as a group (more than one alloy type in the test facility at any one time, possibly including alloys not part of the interlaboratory comparison tests; CIEMAT, Hitachi, JRC, UNB, VTT).

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model

ARTICLE IN PRESS

CS-6637; No. of Pages 10

D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

4

Fig. 2. Photographs of the corrosion coupons before exposure to SCW (Hitachi specimens). Scale is in mm.

Table 5 Descaling solution composition (wt.% in H2 O).

Table 2 Coupon numbering scheme used for the round-robin corrosion tests. Signatory

Laboratory

Coupon identification

Component Solution 1 Citric acid Dibasic ammonium citrate Disodium EDTA Solution 2 Potassium permanganate Caustic soda

08H18Ni10T

800H

310S

Japan

Hitachi

E01 E02 E03

C01 C02 C03

JJ05 JJ06 JJ07

Canada

CNL

E04 E05 E06

C04 C05 C06

J08 J09 J10

CMAT

E07 E08 E09

C07 C08 C09

J11 J12 J13

UNB

E10 E11 –

C10 C11 C12

J14 J15 J16

JRC

E13 E14 E15 E16

C13 C14 C15 C16

J17 J18 J19

VTT

E17 E18 –

C17 – C19

– – –

CIEMAT

E19 E20 E21

– – –

– – –

EU

Table 3 Final parameters for the round-robin tests. In the tests performed in static autoclaves the values for dissolved oxygen, pH and water resistivity represent the initial values. Parameter

Value

Temperature (◦ C) Pressure (MPa) Dissolved oxygen (mg/kg) pH Water resistivity at room temperature (M cm) Test duration (h)

550 25 8 ∼7 at room temperature 18 500

Composition (%) 2 5 0.5 10 4

2.3. Coupon descaling Descaling was performed using a hot alkaline permanganate solution (Solution 2, Table 5) to oxidize Cr(III) to Cr(VI) in Cr-rich oxides and a hot citric acid solution (Solution 1, Table 5) to dissolve the remaining oxides. The coupons were exposed to stirred Solution 1 for 60 min at 90 ◦ C, followed by 60 min in Solution 2 at 90 ◦ C, then exposures to fresh Solution 1 for 30 min at a time until the coupon mass change was constant. Between each step the coupons were removed, sonicated in methanol for 15 min, dried and then weighed. A coupon ground to the same finish but not exposed to SCW was descaled as blanks. The weight was plotted against descaling time; the data showed a distinct change in slope corresponding to the time at which all the oxide was removed and base material corrosion started. The intersection of linear fits to the data before and after the slope change was taken as the sample weight loss. The weight loss of the blank sample was added to this value to obtain the true weight loss. 3. Results 3.1. Weight change The coupon dimension and weight data reported by the interlaboratory comparison participants are listed in Tables 6, 7 and 8 for 800H, 310 SS, and 08H18Ni10T, respectively, and the calculated weight changes are plotted in Figs. 3, 4, and 5 respectively. The coupons showed an increase in weight after exposure to SCW in

Table 4 Main operating parameters of the test facilities used in the round-robin tests. Institute

Type

Flow rate (L/h)

Refresh time (h)

Volume (L)

Volume/coupon surface area (dm)

Autoclave materials

Reference

Hitachi CNL CMAT UNB JRC VTT CIEMAT

Loop Static Static Loop Loop Loop Loop

6 0 0 6 3 0.3 4

0.091 0 0 0.33 0.25 2 1

0.55 0.5 0.5 2 0.8 0.62 4

3.3 3.3 5 1.0 1.95 27

Alloy 625 Alloy 625

[15] [20] [16] [17] [18] [19]

Nimonic 80A Nimonic 80A, Hastelloy C276 Alloy 625

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model

ARTICLE IN PRESS

CS-6637; No. of Pages 10

D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

5

Table 6 Raw data and calculated weight changes for Alloy 800H. Lab

Coupon identification

Dimensions (mm)

L

W

H

Initial mass Surface area (dm2 )(average) (g)

Final mass (average) (g)

Weight change (g)Weight change (mg/dm2 )

Hitachi

C01 C02 C03

19.98 19.98 19.97

9.94 9.96 9.95

1.99 2.00 1.99

0.05225 0.05240 0.05227

2.76842 2.76872 2.77064

2.76861 2.76887 2.77078

0.00019 0.00015 0.00014

3.64 2.86 2.68

CNL

C04 C05 C06

19.937 19.928 19.909

9.847 9.776 9.914

1.939 1.921 1.920

0.0514 0.0510 0.0515

2.95758 2.85309 2.90857

2.96477 2.85908 2.91342

0.00719 0.00599 0.00486

140.01 117.58 94.27

CMAT

C07 C08 C09

19.923 19.939 19.963

9.931 9.914 9.905

1.948 1.939 1.956

0.05180 0.05170 0.05183

2.96431 2.95412 2.97144

2.96461 – 2.97176

0.00030 – 0.00033

5.79 – 6.32

UNB

C10 C11 C12

19.37 19.69 19.73

9.79 9.84 9.86

1.93 1.95 1.93

0.04977 0.05086 0.05091

2.8340 2.9346 2.9434

2.8343 2.9351 2.9444

0.0004 0.0008 0.0010

7.64 15.73 19.64

JRC

C13 C14 C15 C16

19.882 19.910 19.920 19.933

9.905 9.875 9.835 9.856

1.934 1.952 1.956 1.971

0.05149 0.05155 0.05142 0.05164

2.95726 2.98076 2.96705 3.01138

2.95720 2.98098 2.96730 3.01136

−0.00006 0.00022 0.00025 −0.00002

−1.10 4.20 4.93 −0.32

VTT

C17 C19

19.84 19.84

9.90 9.94

1.96 1.96

0.05154 0.05172

2.9862 2.9614

2.9898 2.9627

0.0035 0.0013

68.55 25.14

Table 7 Raw data and calculated weight changes for 08H18Ni10T. Surface area (dm2 )

Initial mass (average) (g)

Final mass (average) (g)

Weight change Weight change (g) (mg/dm2 )

2.00 2.00 2.00

0.05115 0.05098 0.05094

2.79356 2.7863 2.76845

2.79478 2.78764 2.7702

0.00122 0.00134 0.00175

9.608 9.558 9.776

1.877 1.833 1.822

0.04664 0.04559 0.04671

2.35937 2.20752 2.25877

2.37366 2.21713 2.27427

0.01429 0.00960 0.01549

20.131 19.872 19.976

9.96 9.814 9.976

2.00 1.948 1.944

0.05084 0.04918 0.05011

2.72877 2.58071 2.63877

2.72911 2.58104 2.63900

0.00033 0.00032 0.00023

E10 E11

19.85 19.84

9.87 9.91

1.83 1.91

0.04847 0.04924

2.4569 2.5664

2.4570 2.5660

0.0001 0.0014

2.06 28.43

JRC

E13 E14 E15 E16

19.869 19.934 19.960 19.939

9.954 9.960 9.931 9.920

1.934 1.914 1.929 1.922

0.04968 0.04971 0.04976 0.04961

2.78132 2.77858 2.77583 2.78374

2.78250 2.78044 2.77913 2.78672

0.001187 0.001863 0.003307 0.002973

23.89 37.49 66.46 59.94

VTT

E17 E18

19.89 20.02

10.0 9.96

1.95 1.96

0.05005 0.05027

2.81600 2.85747

2.81677 2.85827

0.000767 0.000800

15.32 15.92

CIEMAT

E19 E20 E21

20.038 19.908 19.977

10.04 10.036 10.035

1.916 1.851 1.888

0.05032 0.04949 0.04994

2.6100 2.4931 2.4805

2.6130 2.4956 2.4831

0.00296 0.00248 0.00260

Lab

Coupon identification

Dimensions (mm)

L

W

H

Hitachi

E01 E02 E03

20.15 20.08 20.06

10.02 10.02 10.02

CNL

E04 E05 E06

19.391 19.17 19.292

CMAT

E07 E08 E09

UNB

almost all tests. However, some of the coupons tested at JRC showed weight losses (two of the four Alloy 800H coupons and all of the 310 SS coupons). Several participants reported evidence for exfoliation of the surface oxide for some of the alloys (for example, see Fig. 6), which will reduce the weight gain measured. This may account for some of the variation between laboratories. Oxide exfoliation from austenitic stainless steels has been well documented [21]. In general, the scatter in the weight change data was small within the same laboratory (typically less than ±20% of the average value). However, there was a large variation in the weight change data between different laboratories for the same material (typically greater than ±100% of the average value). There was general, but not unanimous, agreement between the participants regarding

23.85 26.28 34.36 306.4 210.6 331.8 6.56 6.56 4.65

58.82 50.11 52.07

the relative ranking of the three alloys with respect to corrosion, with 310 SS typically showing the lowest average weight gain and 08H18Ni10T the highest. When the average values of all of the data are considered (Table 9), the ranking of the different alloys is more definitive (in order of highest to lowest weight gain): 08H18Ni10T > Alloy800H > 310SS Assuming that the weight gain correlates directly with corrosion rate, this trend is consistent with the increasing chromium content of the three alloys from 08H18Ni10T to 310 SS; it is well established that the corrosion resistance of austenitic steels in SCW generally increases with increasing Cr content. Given that the tests were carried out using coupons from the same heat, prepared using the same procedure, and tested under

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model

ARTICLE IN PRESS

CS-6637; No. of Pages 10

D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

6

Table 8 Raw data and calculated weight changes for 310SS. Weight changes shaded in gray were from extrapolated to 500 h from a shorter exposure time (∼200 h) assuming a parabolic rate law and are included for completeness. Surface area (dm2 )

Initial mass (average) (g)

Final mass (average) (g)

1.99 1.99 1.99

0.050675 0.050699 0.050621

2.79161 2.76774 2.7727

2.79174 2.7679 2.77288

0.00013 0.00016 0.00018

2.57 3.16 3.56

9.769 9.703 9.826

1.904 1.889 1.886

0.048841 0.048342 0.048827

2.52861 2.50250 2.52742

2.53042 2.50501 2.52970

0.00181 0.00251 0.00227

37.10 51.92 46.57

20.046 20.042 20.048

9.984 9.998 9.991

2.002 2.004 2.003

0.050759 0.050827 0.050803

2.77895 2.77976 2.78295

2.77915 2.77998 2.78312

0.00033 0.00035 0.00027

6.41 6.82 5.25

J14 J15 J16

19.84 19.59 19.56

9.82 9.88 9.84

1.91 1.91 1.92

0.048844 0.048516 0.048350

2.50430 2.54492 2.56906

2.50484 2.54504 2.56950

0.00054 0.00012 0.00044

11.06 2.47 9.10

J17 J18 J19

20.045 20.037 20.046

2.011 2.016 2.014

0.050897 0.050917 0.050927

2.79651 2.80040 2.78497

2.79588 2.79977 2.78440

−0.00063 −0.00063 −0.00057

−12.44 −12.37 −11.19

Lab

Coupon Dimensions (mm) identification

L

W

H

Hitachi

JJ05 JJ06 JJ07

20.02 20.03 20.03

10.00 10.00 10.00

CNL

J08 J09 J10

19.953 19.912 19.9

CMAT

J11 J12 J13

UNB

JRC

10.000 10.000 10.001

160

Weight Change (mg/dm2)

Weight Change (mg/dm2)

300

120 100 80 60 40 20

250 200 150 100 50

C18 C17

C16 C15 C14 C13

C12 C11 C10

C09 C08 C07

C06 C05 C04

C03 C02 C01

E21 E20 E19

E18 E17

nominally the same conditions, the data in Figs. 3, 4 and 5 suggest extremely poor reproducibility between test facilities. For example, the weight gains reported by CNL were consistently higher (by at least an order of magnitude) than the values reported by other laboratories. No obvious systematic error could be found in the procedure followed at CNL. It is well known that the initial surface state (polished, machined, etc.) significantly affects the oxidation of austenitic steels in SCW [22]. The surface finishing up to 1200 grit was performed by each participating institution, which introduced some uncertainty regarding the uniformity of this step (manually, automatically, amount of force used, etc.). Examination of Table 4 shows significant differences between the test facilities that were not controlled (or even recognized) during the development of the test procedures. For example, the CNL tests were performed in static autoclaves, which limit the transport of reactants and products to and from the surface. Thus it is not unexpected that these results might be different from those obtained in a flowing system. However, the weight gains reported by CMAT, also from tests carried out in static autoclaves, were very

E16 E15 E14 E13

Fig. 3. Weight change data for 800H from Table 6. The coupon identification information is given in Table 2. Coupons are grouped by participant; from left to right these are Hitachi, CNL, CMAT, UNB, JRC, VTT.

E11 E10

Coupon Idenficaon

E09 E08 E07

E06 E05 E04

0

E03 E02 E01

-20

Weight change (mg/dm2 )

350

140

0

Weight change (g)

Coupon Idenficaon Fig. 4. Weight change data for 08H18Ni10T from Table 7. The coupon identification information is given in Table 2. Coupons are grouped by participant; from left to right these are Hitachi, CNL, CMAT, UNB, JRC, VTT, CIEMAT.

low. This difference was initially puzzling, but it was discovered that after SCW exposure and prior to weighing, the coupons at CMAT were ultrasonicated, which could have dislodged loose oxide (e.g., loose precipitated layers or nearly exfoliated oxide) leading to a low measured weight gain. This treatment was not performed by other participants as it was not part of the agreed upon procedure.

3.2. Oxide descaling Selected coupons from the tests at CNL and VTT were descaled to determine the metal loss due to corrosion. Typical descaling data are shown in Fig. 7, plotted as weight loss versus descaling time. The curve shows a break when the majority of the oxide layer has been removed. The intercept of the two lines is taken as the weight after oxide removal. The weight losses (WL) calculated from the descaling data are given in Table 10. The weight gain (WG) expected

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model

ARTICLE IN PRESS

CS-6637; No. of Pages 10

D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

7

Table 9 Average value for the weight gain reported by each participant, and average of the averages. The column ‘Average (loops only)’ excludes the data from AECL and CMAT. Alloy

Average weight gain (mg/dm2 )

Cr

08H18Ni10T 800H 310

17–19 22.5 24.5

Hitachi

CNL

CMAT

UNB

JRC

VBT

CIEMAT

28.2 3.06 3.09

271.2 117.4 45.2

5.61 6.05 6.16

15.25 14.3 7.54

46.9 1.93 −12.00

15.62 46.84 –

53.67 – –

Average

Average (loops only)

62.4 ± 93.7 31.6 ± 45.3 10.0 ± 21.2

31.9 ± 17.7 16.5 ± 21.0 −0.46 ± 10.2

Table 10 Weight losses obtained by descaling, and the calculated weight gains. Coupon

08H18Ni10T E05 (CNL) E06 (CNL) E17 (VBT) Alloy 800H C04 (CNL) C05 (CNL) C17 (VBT) 310 SS J08 (CNL) J09 (CNL)

Weight Loss (g)

Weight Loss (mg/dm2 )

Calculated Weight Gain (mg/dm2 )

Measured Weight Gain (mg/dm2 )

Calculated Weight Gain/Measured Weight Gain

0.02252 0.03027 0.0006

493.97 648.05 11.881

187.71 246.26 4.5145

210.63 331.77 15.32

0.89 0.74 0.30

0.00847 0.00709 0.0041

164.79 139.02 80.197

62.619 52.828 30.475

140.01 117.58 68.55

0.45 0.45 0.45

0.00261 0.0023

53.439 47.578

20.307 18.080

37.1 51.92

0.55 0.35

60

Interestingly, the ratios of the calculated to measured weight gains for Alloy 800H and 310 SS were similar (∼0.45), with identical values being obtained by CNL and VTT in spite of the differences in test facility. VTT examined the oxides in cross-section, allowing a comparison of the oxide mass calculated from the thickness measured from the SEM images and the oxide mass obtained from the weight gain and weight loss. Fig. 8 shows a low magnification SEM image of an Alloy 800H coupon cross-section after the SCW exposure. There are several questions that must be addressed when calculating the oxide mass from the SEM cross-sections:

Weight Change (mg/dm2)

50 40 30 20 10

J17

J18

J19

J17

J18

J19

J16

J14

J15

J12

J13

J11

J10

J09

J08

JJ06

JJ07

JJ05

0 -10 -20

Coupon Idenficaon Fig. 5. Weight change data for 310S SS. The coupon identification information is given in Table 2. Coupons are grouped by participant; from left to right these are Hitachi, CNL, CMAT, UNB, JRC. Weight changes shaded in gray were extrapolated to 500 h from a shorter exposure time (∼200 h) assuming a parabolic rate law and are included for completeness.

if the metal making up the oxide layers came entirely from the base metal can be easily calculated:



WG = WL ×



fo/ox



1 − fo/ox



(1)

where fo/ox is the weight fraction of oxygen in the oxide (e.g., for Fe3 O4 , fo/ox = 0.276). Since the molecular masses of Fe, Cr and Ni are similar, the assumption that the oxide is magnetite rather than a specific Fe–Cr spinel introduces only a small error. Examination of the data in Table 10 shows that the weight losses measured for coupons exposed in a static autoclave (CNL) and in a flowing loop (VTT) are different, having roughly the same relationship to each other as the weight gain data. This shows that the much larger weight gains observed for the coupons exposed in the CNL autoclaves are not due to increased deposition of corrosion products in the static system. In fact, the corrosion rate was much higher in the static autoclave than in the loop.

1) What does one take as the oxide thickness if the oxide layer is inhomogeneous? 2) How do you account for the oxide porosity in the calculation? For the Alloy 800H sample in Fig. 8, where the oxide was clearly inhomogeneous, it was assumed that 50% of the surface was covered by a 4-␮m-thick oxide and 50% by a thin, 250 nm oxide. The porosity of the thin oxide was assumed to be zero and the porosity of the thick oxide was assumed to be 50% (similar to oxides found on surfaces in the current generation of water-cooled reactors operated at lower temperatures). Using these assumptions, the total mass of oxide per square decimeter of surface was calculated to be 63 mg/dm2 , in good agreement with the weight gain measured for Alloy 800H coupon C17 of 69 mg/dm2 . The contribution of the thinner oxide to the total mass is fairly small (around 10%). Incomplete removal of this oxide by descaling would lead to a lower apparent weight gain, as observed (the oxide mass calculated from the weight loss data is only 31 mg/dm2 ). The oxide on the 08H18Ni10T coupon was much more uniform in appearance, and, using an average oxide thickness of 0.5 ␮m and a porosity of zero, the calculated oxide mass is 28 mg/dm2 . The measured weight gain for 08H18Ni10T coupon E17 was 15 mg/dm2 , suggesting a porosity of 50% would be more appropriate. The oxide mass calculated from the descaling data was only 5 mg/dm2 . Considering that the oxide thicknesses are based on micro-scale measurements while the weight gain and weight loss are macro-scale measurements, the agreement seems reasonable. It is not clear why the weight gain calculated from the descaled weight loss is only about 50% of the measured weight gain, but this

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model

ARTICLE IN PRESS

CS-6637; No. of Pages 10

D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

8

Fig. 6. SEM surface morphologies of Alloy 800H coupon C19 (top) and 3D profile measurement of the corresponding sample surface (bottom) after 500 h of exposure to SCW at 550 ◦ C, 25 MPa, 8 mg/kg DO2 at VTT, showing evidence of oxide exfoliation.

2.505

Coupon Weight (g)

2.504 2.503 2.502 2.501 2.500 2.499

0

50

100

150

200

250

300

Descaling Time (minutes) Fig. 7. Coupon weight as a function of descaling time for 310 SS coupon J09.

could be due in part to internal oxidation of the base metal; such layers may not be easily removed by the descaling process. It is also possible that some material originating from corrosion of the loop and autoclave alloys precipitated on the coupons. Experiments at CNL suggest that repeated transitions through the critical point, for example due to power supply trips, can lead to higher weight gains than observed in experiments that run continuously with no trips. However, the descaling data presented in the current work suggest that the higher weight gain was associated with a higher corrosion rate, and further work is needed to understand these observations. 3.3. Effect of flow rate The dependence of the weight gain on the various facility parameters listed in Table 6 was examined. Only the autoclave refresh

Fig. 8. Low magnification SEM image of 800H coupon C19 after exposure to SCW for 500 h under the interlaboratory comparison test conditions.

time appeared to show a reasonable correlation with weight gain (Figs. 9 and 10). The autoclave refresh time determines both the transport of oxidants to the surface and the transport of corrosion products away from the surface. Several of the test loops monitored the dissolved oxygen concentration before and after the test section, and could verify that the oxygen concentration at the corrosion coupons met the specification. JRC reported a systematic loss of about 600 ppb between inlet and outlet, assumed to be due to corrosion of the corrosion coupons and also the high temperature sections of the loop, including the autoclave itself. In a static autoclave, the water chemistry cannot be controlled during a test. Oxidant concentrations will decrease with time, and corrosion product concentrations in the test solution can reach sat-

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model

ARTICLE IN PRESS

CS-6637; No. of Pages 10

D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

Weight Gain (mg/dm2)

140

volume, there may be a temperature gradient across the specimen rack, and therefore the location of the samples in the autoclave could be important. JRC suggested that a temperature gradient of as much as 10–15 ◦ C could exist in the autoclave. In a loop, the inlet and outlet water flows into the autoclave can be in different locations (e.g., both at the bottom, both at the top, at opposite ends). In a static autoclave, the coupons are initially in a two-phase system (liquid and vapour), immersed in either the liquid phase or the vapor phase, and could be exposed to a range of conditions during system heat-up and cool-down.

800H

120

310

100 80 60 40 20 0 -20

4. Conclusions 0.01

0.1

1

10

100

1000

Autoclave Refresh Time (h) Fig. 9. Weight gain as a function of autoclave refresh time for Alloy 800H and 310 SS.

300

Weight Gain (mg/dm2)

9

250 200 150 100 50 0 0.01

0.1

1

10

100

1000

Autoclave Refresh Time (h) Fig. 10. Weight gain as a function of autoclave refresh time for 08H18Ni10T.

uration and limit oxide dissolution into the water. In a flow loop, the water chemistry can be easily monitored and controlled, and corrosion products are removed from the vicinity of the surface, allowing dissolution of the surface oxide. The descaling data suggest that the transport of corrosion products in a static autoclave was not a limiting factor, since the weight losses for each alloy tested in the CNL static autoclave were much higher than those tested at VTT in a flowing loop. Oxide exfoliation may be higher at the higher flow rates in loops, resulting in lower weight gains. However, while examination of the coupons at VTT showed evidence for oxide exfoliation, the weight gain calculated from the descaled weight loss was lower than the measured weight gain, which is inconsistent with a high oxide loss by exfoliation. Increased transport of oxygen to the coupon surface in a flow system may lead to increased oxidation of the alloy (e.g., due to loss of the protective Cr-rich oxide film due to formation of soluble Cr(VI) species [23]). Increased oxygen transport may also reduce corrosion by promoting the rapid formation of the protective Crrich oxide film if the concentration of oxygen is below that required for the formation of soluble Cr(VI) species. 3.4. Other factors There are a number of other design differences related to the autoclave or test section holding the corrosion coupons that may also affect the results. For example, depending on the autoclave

As noted in the introduction, it was intended that “. . .roundrobin testing and characterization of identical alloys under similar test conditions be carried out to assess the reproducibility of the results. . .” Such an interlaboratory comparison was considered necessary given the large number of research groups world-wide performing corrosion testing in support of the SCWR concept and the collaborative nature of the GIF. It would be safe to say that the results of the interlaboratory comparison test were surprising to all the participants. Considerably more variation in the weight gain was observed than was expected. For the same material , the scatter in the weight change data was small within the same laboratory (typically less than ±20% of the average value), but large between different laboratories (typically greater than ±100% of the average value). Much of this variation appears to be attributable to differences in the test facilities used; these differences were not considered during the development of the interlaboratory comparison test procedures. The autoclave refresh time (flow rate) appears to play a large role, although the specific mechanism(s) could not be determined from the available data. In spite of these variations, the data do generally agree on the relative ranking of the corrosion resistance of the three alloys tested; taken together, the dataset is more definitive, ranking the alloys (in order of highest to lowest weight gain) as follows: 08H18Ni10T > Alloy800H > 310SS This trend is consistent with the increasing chromium content of the three alloys. Several conclusions can be drawn from the results of the roundrobin corrosion tests: 1. Weight gain data must be interpreted carefully, and in themselves are insufficient to quantify the corrosion rate of materials in SCW. 2. Care must be taken when comparing data obtained in different test facilities. Details of the facilities and the experimental pre/post-treatments must be reported and taken into consideration when assessing the data. 3. The relative ranking of materials with respect to corrosion susceptibility obtained from weight gain data is reliable. 4. Development of standard test methodologies for corrosion tests in support of SCWR development is recommended. The GIF SCWR M&C PMB is planning a second phase of interlaboratory comparison corrosion tests to address some of the deficiencies identified in the first phase. In this second phase, only one alloy type will be tested and all of the coupons will be prepared (including surface finishing) by one institute. Key loop parameters will be identified and, to the extent possible, made as uniform as possible (e.g., autoclave refresh time).

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034

G Model CS-6637; No. of Pages 10 10

ARTICLE IN PRESS D. Guzonas et al. / Corrosion Science xxx (2016) xxx–xxx

Acknowledgements This work was carried out as part of a joint activity of the Generation IV International Forum SCWR Materials and Chemistry Project Management Board, with contributions from Canada, Euratom and Japan. For brevity, only the lead author from each participating institution appeared in the list of authors. The authors are grateful to many others who participated in these tests, and would like to acknowl˜ (CIEMAT); J. edge the following individuals: D. Gómez-Briceno Collier and M. Matchim (CMAT); L. Deschenes, M. Edwards, S. Rousseau (CNL). The financial support of Academy of Finland project IDEA (Interactive modelling of fuel cladding degradation mechanism) to VTT is gratefully acknowledged. CNL and CMAT acknowledge the financial support of the Natural Resources Canada Office of Energy Research and Development. UNB acknowledges the financial support of the NSERC/NRCan/AECL Generation IV Technologies program. References [1] B. Cox, Accelerated Oxidation of Zircaloy-2 in Supercritical Steam, Atomic Energy of Canada Limited Report AECL-4448, 1973. [2] A.T. Motta, A. Yilmazbayhan, M.J. Gomes da Silva, R.J. Comstock, G.S. Was, J.T. Busby, E. Gartner, Q. Peng, Y.H. Jeong, J.Y. Park, Zirconium alloys for supercritical water reactor applications: challenges and possibilities, J. Nucl. Mater. 371 (2007) 61–75. [3] D. Khatamian, Corrosion and deuterium uptake of Zr-based alloys in supercritical water, J. Supercrit. Fluids 78 (2013) 132–142. [4] T.R. Allen, Y. Chen, X. Ren, K. Sridharan, L. Tan, G.S. Was, E. West, D.A. Guzonas, Material performance in supercritical water, in: R.J.M. Konings (Ed.), Comprehensive Nuclear Materials, vol. 5, Elsevier, Amsterdam, pp. 279–326. [5] R. Peraldi, B.A. Pint, Effect of Cr and Ni contents on the oxidation behavior of ferritic and austentitic model alloys in air with water vapor, Oxid. Met. 61 (2004) 463–483. [6] F.A. Comprelli, D.F. MacMillan, C.N. Spalaris, Materials for superheated fuel sheaths: relative performance of alloys in superheated steam environments, General Electric Company Report GEAP-4351, 1963. [7] Ya. Emel’yanov, O.A. Shatskaya, E.Yu. Rivkin, N.Ya. Nikolenko, Strength of Construction Elements in the Fuel Channels of the Beloyarsk Power Station Reactors Atomnaya Energiya, 33 (1972) 729–733 (in Russian) (Translated in Soviet Atomic Energy 33 (3) (1972) pp. 842–847). [8] P.B. Longton, The Oxidation of Iron- and Nickel-based Alloys in Supercritical Steam: A Review of the Available Data, UKAEA The Reactor Group, TRG Report 1144 (C), 1966.

[9] X. Ru, R.W. Staehle, Historical experience providing bases for predicting corrosion and stress corrosion in emerging supercritical water nuclear technology: part 1—review, Corrosion 69 (2013) 211–229. [10] I.G. Wright, R.B. Dooley, A review of the oxidation behaviour of structural alloys in steam, Int. Mater. Rev. 55 (2010) 129–167. [11] R. Viswanathan, J.F. Henry, J. Tanzosh, G. Stanko, J. Shingledecker, B. Vitalis, R. Purgert, U.S. program on materials technology for ultra-supercritical coal power plants, J. Mater. Eng. Perform. 22 (2013) 2904–2915. [12] D. Guzonas, SCWR Materials and chemistry—status of ongoing research, Proceedings of the GIF Symposium, Paris, France, September 9–10, 2009, 163–172. [13] D. Guzonas, F. Brosseau, P. Tremaine, J. Meesungnoen, J.-P. Jay-Gerin, Water chemistry in a supercritical water-cooled pressure tube reactor, Nucl. Technol. 179 (2012) 205–219. [14] V. Subramanian, J.M. Joseph, H. Subramanian, J.J. Noël, D.A. Guzonas, J.C. Wren, Steady-state radiolysis of supercritical water: model development, predictions and validation, 7th Int. Symposium on Supercritical Water-Cooled Reactors (ISSCWR-7), Helsinki, Finland, 15–18 March, 2015, Paper ISSCWR7-2083. [15] D.A. Guzonas, J. Wills, T. Do, J. Michel, Corrosion of candidate materials for use in a supercritical water CANDU® reactor, 13th International Conference on Environmental Degradation of Materials in Nuclear Power Systems (2007) 1250–1261. [16] W. Cook, J. Miles, J. Li, S. Kuyucak, W. Zheng, Preliminary analysis of candidate alloys for use in the CANDU-SCWR, The 2nd Canada–China Joint Workshop on Supercritical Water-Cooled Reactors (CCSC-2010), April 25–28, 2010. ´ P. Janík, S. Penttilä, P. Hähner, J. Macák, J. Siegl, P. Hauˇsild, High Cr [17] R. Novotny, ODS steels performance under supercritical water environment, J. Supercrit. Fluids 81 (2013) 147–156. [18] S. Penttilä, A. Toivonen, L. Rissanen, L. Heikinheimo, Generation IV material issues—case SCWR, J. Disaster Res. 5 (2010) 469–478. ˜ F. Blázquez, A. Sáez-Maderuelo, Oxidation of austenitic [19] D. Gómez-Briceno, and ferritic/martensitic alloys in supercritical water, J. Supercrit. Fluids 78 (2013) 103–113. [20] J.L. Krausher, W. Zheng, J. Li, D. Guzonas, G. Botton, Filling the gap in SCWR materials research: advanced nuclear corrosion research facilities in Hamilton, 32nd Annual Canadian Nuclear Society Conference, Niagara Falls, ON, June 5–8, 2011. [21] I.G. Wright, P.F. Tortorelli, M. Schütze, Program on Technology Innovation: Oxide Growth and Exfoliation on Alloys Exposed to Steam, Electric Power Research Institute Report 1013666, 2007. [22] W.E. Ruther, R.R. Schlueter, R.H. Lee, R.K. Hart, Corrosion behavior of steels and nickel alloys in superheated steam, Corrosion 22 (1966) 147–155. [23] O.S. Bakai , D.A. Guzonas , V.M. Boriskin , A.M. Dovbnya , S.V. Dyuldya , Combined effect of irradiation, temperature, and water coolant flow on corrosion of Zr-, Ni–Cr-, and Fe–Cr-based alloys, The 7th International Symposium on Supercritical Water-Cooled Reactors (ISSCWR-7), Helsinki, Finland, 15–18 March, 2015, Paper ISSCWR7-2012.

Please cite this article in press as: D. Guzonas, et al., The reproducibility of corrosion testing in supercritical water—Results of an international interlaboratory comparison exercise, Corros. Sci. (2016), http://dx.doi.org/10.1016/j.corsci.2016.01.034