Doel 3 & Tihange 2 - Some Peer-reviewed Scientific Papers & Reports

Hydrogen Cracks in Doel 3 & Tihange 2

Some Peer-reviewed Scientific & Technical Papers or Study Reports …

Hydrogen-induced Damage in PWR Reactor Pressure Vessels

W.F. Bogaerts (1) , J.H. Zheng (2) , A.S. Jovanovic (3) & D.D. Macdonald (4)

(1) University of Leuven – Dept. of Materials Engineering & Center for Nuclear Technology (Belgium) (2) Technologica Group – INS, Intelligent Nuclear Solutions Div. (Belgium) (3) Eu-VRi – European Virtual Institute for Integrated Risk Management (Germany) (4) University of California, Berkeley – Dept. of Nuclear Engineering, Berkeley, CA 94720 (USA) Recently, the potential problem of (hydrogen-related) cracks in RPV-steels has become imminent in the Belgian nuclear power reactors Doel 3 and Tihange 2. This communication briefly elaborates on some reported findings and identifies possible mechanisms for the detected flaws in the reactor pressure vessel walls. Stress corrosion cracking, pitting corrosion, and other materials degradation mechanisms in reactor coolant system components have cost the nuclear industry billions of dollars, due to forced and extended outages, component repair and replacement, and increased inspection requirements and regulatory scrutiny. Materials aging effects must be effectively managed to ensure that safe and reliable functionality is maintained throughout the life of the plant, especially if life extension of nuclear power plants (NPP) is contemplated. In this respect, specific R&D activities on plant integrity and materials reliability have been launched in recent years 1 . Since mid-2012, however, the situation for a number of PWR plants may have dramatically changed. Ultrasonic inspections of the Reactor Pressure Vessels (RPV) at the Belgian PWR nuclear power plants Doel 3 and Tihange 2 (1000 MWe), which started operation in 1982-1983, have revealed a large number of (small) cracks in the RPV wall 2 , causing the operator and the national regulatory authority FANC (Federal Agency for Nuclear Control) to shut down the reactors and commence further testing. Other plants around the world, both in the US and Europe, are operating reactor pressure vessels manufactured by the same company or using steels from the same producer (at least 22 in total). Introduction

Figure 1: Illustration translated from FANC, showing the original forged steel ring sections of the RPV separated for clarity. These rings are welded together and cladded internally with a stainless steel lining to form the reactor pressure vessel.

1

Experimental Findings

Figure 1 shows a schematic illustration of the affected RPVs. The total height of the vessel is approx. 13 meters (incl. the spherical top lid), with a diameter of 4.4 meters and a wall thickness of the cylindrical part of 205 mm. The primary water side of the RPV is cladded with a stainless steel Type 304 lining of approx. 7 mm thickness. In June 2012 ultrasonic in-service inspections, using a new technique, were performed at the Doel 3 NPP, in order to check for underclad cracking in the reactor pressure vessel, as had been found at Tricastin (France). No such defects were detected. However, unexpected atypical “indications” in the RPV shells were found in the first 30 mm of the material depth of the irradiated part of the Doel 3 RPV core shells. Hence, the operator ordered a full thickness RPV shell inspection in July 2012, which confirmed high numbers (thousands) of similar “indications” 3 down to a depth of 120 mm into the material, measuring from the reactor’s primary water side . It appeared that flaws were particularly dominant in the bottommost and upper core shells. The bulk of them are located in the base metal, outside the weld regions. Flaw densities as high as 40 indications per dm 3 have been found, with a reported total of 7776 indications in the core lower shell (core upper shell showed 931 indications). The flaws appeared “almost circular in shape” w ith an average diameter of 10-14 mm, although some had diameters as large as more than 20-25 mm. It was also observed that the detected defects are oriented laminar or quasi-laminar 3-5 and that their position and orientation showed a pattern similar to the pattern of a zone of macro-segregations. Bridging was found to occur only between flakes that are very close to each other. At that time, FANC apparently also declared that similar UT inspections of the RPV head and upper rings in the 1990s found only a few indications 2 . Old UT-inspection records, dating from the time of fabrication of the forgings, also did not mention the significant presence of “indications”. In September 2012 the same type of “ indications ” , but to a lesser extent, was also found in the Tihange 2 RPV shells during a similar inspection. Both RPV forgings were produced by the same fabricator with steel from the same supplier. Different investigations have been carried out since the flaws were first discovered 3-4 . They have highlighted so-called 'hydrogen flakes' as being the root cause of the problem. These hydrogen flakes might arise during the fabrication of large steel ingots 6 . Solidification of a large mass of steel is characterized by significant development of micro- and macro-defects in the ingot structure and a changing solubility of different elements during cooling. For example the solubility of hydrogen (e.g. originating from thermal dissociation of water molecules from damp scrap, fluxes, atmospheric humidity, etc.) decreases during solidification and cooling down of the steel ingot. The solubility of hydrogen in steel at room temperature is approx. 0.1 ppm, compared to 30 ppm in the steel melt. Hydrogen atoms possess a high mobility in the steel matrix, but are collected at internal voids, such as non-metallic inclusions (sulfides, oxides), shrinkage pores, cracks caused by internal stresses, etc. Hydrogen atoms collected at such internal micro-voids combine and form gaseous hydrogen molecules H 2 , which may cause formation of cracks (“flakes” , in the traditional steel jargon) when the gas pressure exceeds the steel strength. Hydrogen flakes (sometimes also called ‘ shatter cracks ’: Discussion Causes of the Cracks

2

internal fissures seen in large forgings due to segregated hydrogen) were well known from the past and their possible formation is particular dangerous for parts fabricated from large ingots. A potential remedy is to use vacuum ladle degassing methods to decrease the content of hydrogen to 2 ppm, which should avoid or mitigate flake formation. Not all forged components of the Doel 3 and Tihange 2 RPVs contain the same amount of flaws. Based on an analysis of the ingot size and the combined sulfur and hydrogen content, there appears to be a good correlation with the intrinsic susceptibility to hydrogen flaking and the amount of flakes found in each forged component. The key question remaining is about the possible evolution over time of these so- called “hydrogen flakes”. The position of the regulatory authorities and the operator, so far, has been that the defects found in the Doel and Tihange RPV “are usually associated with manufacturing and are not due to aging” and that it is “improbable” that the flaws have evolved since their formation. The only theoretical propagation mechanism still considered is ‘low cycle fatigue’ 4 , while also the limited experience about the influence of irradiation on flaw propagation in zones with hydrogen flakes is recognized. This phenomenon is currently under investigation. One of the main reasons for concluding that it is unlikely there has been a significant evolution of the voids over time is the claim that “there is currently no source of hydrogen anymore” which could cause propagation of the cracks. This, however, is an erroneous conclusion. The phenomenon described above is very reminiscent of well- known ‘hydrogen blistering’ or hydrogen induced fracture phenomena from corrosion in the chemical and petrochemical industries. Hydrogen blistering can occur when hydrogen enters steels as a result of the reduction reaction (hydrogen evolution via water and/or proton reduction) on a corroding metal surface. In this process, single-atoms of “ nascent ” hydrogen diffuse through the metal until they react with another atom, usually at inclusions or defects in the metal. The resultant diatomic hydrogen molecules are then too large to migrate through the metal lattice and become trapped. Eventually, a gas blister or internal crack builds up and may split the metal as schematically illustrated in Figure 2. Practical examples are shown in Figure 3. Water chemistry, corrosion effects and hydrogen

Figure 2: Schematic diagram of hydrogen diffusion and blister formation.

3

Figure 3: Typical hydrogen induced cracks (source: MTI Atlas of Corrosion and Related Materials Failures – electronic ed.)

Hydrogen blistering or cracking is controlled by minimizing corrosion and is normally not a problem in neutral or alkaline environments and with high-quality steels that have low impurity and inclusion levels. Nevertheless also under the primary water chemistry conditions of the reactor coolant system (RCS) of PWRs, with a typical pH T of approx. 6.9 to 7.4 (corresponding to a room temperature pH around 10; cf. Figure 4) the primary cathodic corrosion reaction will be: - → H + OH - Even for very low corrosion rates of the stainless steel cladding (e.g. 0.1 to 1 micron/yr) this will result in significant quantities of corrosion-generated hydrogen atoms that may enter into the base metal. H 2 O + e

Figure 4: pH control in PWR primary coolant by adjusting the lithium concentration as the boron is consumed during fuel burn-up. The trajectory commonly employed over a typical fuel cycle is marked by the dark path (EPRI PWR Primary Water Chemistry Guidelines TR-105714-V1R4).

The austenitic stainless steel cladding is sometimes considered to prevent hydrogen diffusion and potential hydrogen-induced cracking problems in the pressure vessels. This, to our knowledge, has never been proven experimentally in an adequate way 7 and, at most, the cladding probably has only a “delaying” effect in transferring the nascent hydrogen to the cladding/base metal boundary, and further into the RPV steel matrix. The presence of flaws in this matrix (cf. “hydrogen flakes”) represents ideal sinks (traps) for the hydrogen injected into the metal from the cathodic corrosion reaction. In addition to the corrosion-generated hydrogen, there is also the issue of hydrogen radicals being formed as a result of the radiolysis of water and the reactions of H 2 with the radiolysis products 8 (e.g. OH ⋅ + H2 → H ⋅ + H2O); hydrogen is used in the RCS to suppress radiolytic oxygen and hydrogen peroxide formation. More details of all these effects are described elsewhere 9 and will be discussed

4

during the meeting.

Another argument relates to the size of the observed defects. Typical void sizes of hydrogen flakes are generally reported to be 10 to 15 mm, while current inspections have found sizes up to 25 or 30 mm. Finally, there are earlier observations of hydrogen-induced blister cracking which have been reported in nuclear structural materials 10 , and in the past there has been a lot of debate about the issue. A very old, specific, example of failures attributed to hydrogen occurred in retaining rings used to connect inlet assemblies to the reactor process tubes in a Hanford water cooled production reactor. Failures occurred in carbon steels and Type 420 stainless steel. The reported hydrogen sources were the fabrication process, hydrogen generated during corrosion of the ring by the process water, and from galvanic coupling 11 . In view of all of the above, the “trapping” of cathodically generated hydrogen (due to primary water corrosion reactions) inside existing “hydrogen flakes” is not improbable. Moreover, the (original) flakes may act as a stress raiser, which will enhance the diffusion of the hydrogen to the stressed areas in the metal. Also the additional effect of irradiation is still largely unknown. After some 2 years of investigation, it remains unclear if the cracks found in the Belgian NPPs Doel 3 and Tihange 2 are "only" manufacturing artefacts, or if there is also an "operational component" contributing to the current problems and operational risks; i.e. whether the cracks are still progressing and whether there are other phenomena, e.g. similar to 'hydrogen blistering' processes, contributing to the problem. Additional hydrogen might indeed come from the cathodic corrosion reactions occurring on the primary water side of the reactor pressure vessel. During operation, there is a permanent flux of (corrosion-originating) atomic hydrogen – although the flux might be small – and this hydrogen could easily get trapped into the voids that are present in the wall of the RPV. An eventual pressure build-up in the flakes will result in growing cracks and other materials degradation phenomena. It is also not just Doel 3 and Tihange 2 in Belgium that could be affected. The RPVs were fabricated by the, now bankrupt, RDM (Rotterdamsche Droogdok Maatschappij, Netherlands), which also manufactured RPVs for at least 20 other reactors that are operating in seven countries around the world, including some 10 in the United States. Of course, also other factors like steel supplier, cladding process and final assembling may have played an important role in the development of the observed damage. If some of the initial hypotheses discussed above were to be proved to be true, there might be a huge impact on currently operating PWRs. Conclusions 1. A. Demma. Pressurized water reactor materials reliability program . EPRI Research Portfolio – Program Overview (2012). 2. C. Peachey. Cracks found at Doel 3 . Nuclear Engineering International, p. 10, October 2012. 3. N.N. (AIB-Vinçotte). Synthesis Report Doel 165 (Internal report), 10 pp. (January 2013). 4. N.N. Doel 3 and Tihange 2 reactor pressure vessels – Final Evaluation Report. FANC (Federal Agency Nuclear Control, Belgium), 33 pp. (May 2013). 5. N.N. Defects in the reactor pressure vessels of Doel 3 and Tihange 2 . The Greens / European Free Alliance (European Parliament), 41 pp. (March 2014). References

5

6. R.J. Fruehan. A review of hydrogen flaking and its prevention . Iron & steelmaker, Vol. 24, Nr. 8, p. 61 (1997). 7. R.E. Mazel, V.G. Grinenko & T.P. Kuznetsova. 1980. Influence of cladding on the resistance to hydrogen embrittlement of the metal of a nuclear reactor vessel. Teploenergetika, Vol. 27, Nr. 10, p. 29 (1980). 8. D.D. Macdonald & M. Urquidi-Macdonald. The Electrochemistry of Nuclear Reactor Coolant Circuits , Encyclopedia of Electrochemistry, Vol. 5 Electrochemical Engineering, Eds. D.D. Macdonald and P. Schmuki, Wiley-VCH Verlag, pp. 665-720 (2007). 9. W.F. Bogaerts & D.D. Macdonald. Hydrogen-assisted cracking in nuclear power plants? J. Nucl. Materials, submitted (2015). 10. T.K.G. Namboodhiri. Trans. Indian Inst. Metals, Vol. 37, p. 764 (1984). 11. R.E. Westerman. An analysis of the Truarc Ring Failure Problem , HW-71631, November 8, 1961; cited in: S.H. Bush & R.L. Dillon , ‘Stress Corrosion in Nuclear Systems’ , Proc. Int. Conf. on Stress Corrosion Cracking and Hydrogen Embrittlement of Iron Base Alloys (NACE-5), p. 61, June 1973.

6

Hydrogen Crac k s in Belgian Nuclear Reactor Pressure Vessels: Five years after their discovery – An update

W.F. Bogaerts University of Leuven (KU Leuven) – Dept. MTM & Center for Nuclear Engineering

A.S. Jovanovic Eu-VRi – European Virtual Institute for Integrated Risk Management Stuttgart, Germany

Kasteelpark Arenberg 44 B-3001 Leuven, Belgium Email: walter.bogaerts@kuleuven.be

J.H. Zheng Technologica Group – INS Div. B-2450 Meerhout, Belgium

D.D. Macdonald University of California, Berkeley – Nuclear Engineering & Materials Science and Engineering Berkeley, USA

K. Dockx University of Leuven (KU Leuven) – Dept. MTM

Kasteelpark Arenberg 44 B-3001 Leuven, Belgium

ABSTRACT

More than three years after the first detection of thousands of “hydrogen flaws” in its Reactor Pressure Vessel (RPV), the Belgian nuclear power reactors Doel 3 and Tihange 2 were restarted at the end of 2015. However – now again almost two years later – the potential problem of (hydrogen-related) cracks in the RPV and the related longer-term aging problems of the reactor are still imminent, probably more than ever. This paper briefly elaborates on some reported findings and identifies possible mechanisms and risks for further growth of these defects in the reactor pressure vessel wall. The current study shows that – despite a number of counter-arguments – there are, for instance, significant potential risks or uncertainties about process-generated hydrogen problems (e.g. originating from corrosion, radiolysis, etc). These are sufficiently grave to raise concerns about the fitness-for-service of the affected reactors, and also about similar reactors world-wide. Just one example of the latter is the finding of similar “UST-indications” in the Swiss nuclear power plants Beznau 1 (and 2). From a life-management and safety point of view, it is recommended that meticulous inspection and continuous monitoring and surveillance programs be set up and implemented when keeping these aging reactors in operation.

Key words: nuclear, RPV, cracks, hydrogen

INTRODUCTION

The inner surface of most western PWR reactor pressure vessels (RPV) is clad with austenitic stainless steel, in order to reduce corrosion rates and CRUD formation. Some of instances, this clad has been reported to develop cracks, which will create direct contact of the base RPV metal (a ferritic low-alloy steel) with the high-temperature coolant (boric acid + lithium hydroxide + hydrogen) in the primary circuit and possible progression of the cracks into the base material. In June 2012, ultrasonic, in-service inspections (UST), using a new technique/instrumentation, were performed at the Belgian Doel 3 nuclear power plant, in order to check for under-clad cracking in the reactor pressure vessel, as had e.g. been found earlier at Tricastin 1 in France. Figure 1 shows a schematic illustration of the Doel 3 RPV. The total height of the vessel is approx. 13 meters (incl. the spherical top lid), with a diameter of 4.4 meters and a wall thickness of the cylindrical part of 205 mm. The primary water side of the RPV is cladded with a stainless steel Type 308/309 lining of approx. 7 mm thickness. The RPV base material is a SA 508 Cl. 3 low-alloy Mn-Mo-Ni steel (i.e. 1.2-1.5% Mn, 0.45-0.60% Mo, 0.40-1.00% Ni, max. 0.25% Cr, max. 0.25% C).

Figure 1: Illustration translated from FANC, showing the original forged steel ring sections of the RPV separated for clarity. These rings are welded together and clad internally with a stainless steel liner to form the reactor pressure vessel.

EXPERIMENTAL FINDINGS

No under-clad cracking defects were detected. However, unexpectedly, a-typical UST- “indications” in the RPV shells were found in the first 30 mm of the material depth of the irradiated part of the Doel 3 RPV core shells. Hence, the operator ordered a full thickness RPV shell inspection in July 2012, which confirmed high numbers (thousands) of similar “indications” 1,2,3 down to a depth of 120 mm into the material, measuring from the reactor’s primary water side. It appeared that the flaws were particularly dominant in the bottommost and upper core shells. The bulk of the flaws are located in the base metal, outside the weld regions. Flaw densities as high as 40 indications per dm 3 were found, with an initially reported total of

7776 indications in the core lower shell (core upper shell showed 931 indications).

The flaws appeared “almost circular in shape” with a reported average diameter of 10-14 mm, “although some had diameters as large as more than 20-25 mm” (some available data, however, showed significantly higher values, cf. infra – See Figures 2 and 3). 4,5 It was also observed that the detected defects are oriented laminar or quasi-laminar 2,6,7 and that their position and orientation showed a pattern similar to the pattern of a zone of assumed macro- segregations. 8 Bridging was found to occur only between flakes that are very close to each other. At that time, FANC (national Federal Agency for Nuclear Control) apparently also declared that similar UST inspections of the RPV head and upper rings in the 1990s found only a few indications. 1 Old UST-inspection records, dating from the time of fabrication of the forgings, also did not mention the significant presence of “indications”.

Figure 2: Doel 3 – Size of indications [max (x,y)] versus depth into the RPV steel wall (data 2012). 4

Figure 3: Tihange 2 – Size of indications [max (x,y)] versus depth into the RPV steel wall (data 2012). 5

In September 2012, the same types of “indications”, but fewer in number, were also found in the Tihange 2 RPV shells during a similar inspection 5 (Figure 3). Both RPV forgings were produced by the same fabricator with steel from the same supplier.

DISCUSSION

Causes of the Cracks

Different investigations have been carried out since the flaws were first discovered. e.g. 2, 6, 8, 9 They have highlighted so-called 'hydrogen flakes' as being the root cause of the problem. These hydrogen flakes might arise during the fabrication of large steel ingots. 10 Solidification of a large mass of steel is characterized by significant development of micro- and macro-defects in the ingot structure and a changing solubility of different elements during cooling. For example, the solubility of hydrogen (e.g. originating from thermal dissociation of water molecules from damp scrap, fluxes, atmospheric humidity, etc.) decreases during solidification and cooling down of the steel ingot. The solubility of hydrogen in steel at room temperature is approx. 0.1 ppm, compared with 30 ppm in the steel melt. Hydrogen atoms possess a high mobility in the steel matrix, but are collected at internal voids, such as non-metallic inclusions (sulfides, oxides), shrinkage pores, cracks caused by internal stresses, etc. Hydrogen atoms collected at such internal micro-voids combine and form gaseous hydrogen molecules H 2 , which may cause formation of cracks (“flakes”, in the traditional steel jargon) when the gas pressure exceeds the steel strength. In fact, the “hydrogen flakes” are internal cracks extending radially in all directions from a center (e.g. inclusion), with the typical characteristics of a hydrogen-induced (transcrystalline) brittle fracture. Hydrogen flakes (sometimes also called ‘ shatter cracks ’: internal fissures seen in large forgings due to segregated hydrogen) are well-known from the past and their possible formation is particular dangerous for parts fabricated from large ingots. A potential remedy is to use vacuum ladle degassing methods to decrease the content of hydrogen to 2 ppm, which should avoid or mitigate flake formation, but apparently this was not done in preparing the ingots from which the RPVs for the Doel and Tihange PWRs were fabricated. Not all forged components of the Doel 3 and Tihange 2 RPVs contain the same amount of flaws. Based on an analysis of the ingot size and the combined sulfur and hydrogen content, there appears to be a good correlation with the intrinsic susceptibility to hydrogen flaking and the amount of flakes found in each forged component. The key question remaining is about the possible evolution over time of these so-called “hydrogen flakes”. The position of the regulatory authorities and the operator, so far, has been that the defects found in the Doel and Tihange RPV “are usually associated with manufacturing and are not due to aging” and that it is “improbable” that the flaws have evolved since their formation. The only theoretical propagation mechanism still considered is ‘low cycle fatigue’. 6 Also the limited experience concerning the influence of irradiation on flaw propagation in zones with hydrogen flakes is, however, recognized. One of the main reasons for concluding that it is unlikely there has been a significant evolution of the voids over time is the claim that “there is currently no source of hydrogen anymore” which could cause propagation of the cracks. Recognizing that the inner surface of the RPV is in

contact with an aqueous solution (primary coolant), this is clearly an incredible and erroneous conclusion.

Hydrogen segregation during the steel production and manufacturing process as being the only cause of the current cracks is also put to discussion by e.g. Boonen et al. 11,12 The so-called ‘Safety Case Report’ for Doel 3 states that the concentration of hydrogen decreases from 1.5 to 0.8 ppm during cooling of the steel. A first estimate for the high-density flaw region of the lower shell shows that this would equal to about 61 mL H 2 in total. In order to obtain an approximation of the amount of hydrogen needed to generate all the flaws in this section of the RPV, Boonen et al. have carried out a number of linear elastic finite element simulations to estimate the total flaw volume. Based on this theoretical calculation, they come up with a totally needed H 2 volume of 604 mL, which is approximately 10 times the stated released amount of hydrogen from the steel. These calculations appear to be quite realistic, since they would also result in a calculated average flaw/crack opening width of 0.25 to a few micrometer. The conclusion from this study is that traditional “hydrogen flaking” cannot be the only cause of the present flaws. Either another cause has resulted in the flaking, or the flaws have been growing over time. Boonen et al. 11,12 have also severely questioned the validity of the fracture mechanics approach used by the operator. Interaction of the many multiple cracks as seen in the current case is not covered by the ASME approach. The flaking phenomenon described above is very reminiscent of the well-known ‘hydrogen blistering’, ‘water blistering’ or hydrogen-induced fracture phenomena from corrosion in the chemical and petrochemical industries. Hydrogen blistering can occur when hydrogen enters steels as a result of the reduction reaction (hydrogen evolution via water and/or proton reduction) on a corroding metal surface. In this process, single-atoms of “nascent” hydrogen (H atoms) diffuse through the metal until they react with another atom, usually at inclusions or defects in the metal. The resultant diatomic hydrogen molecules are then too large to migrate through the metal lattice and become trapped. Eventually, a gas blister or internal crack builds up and may split the metal as schematically illustrated in Figure 4. Practical examples are shown in Figure 5. In the presence of already existing cracks (cf. the “hydrogen flakes”) the newly generated hydrogen may be responsible for further crack growth in two ways: either through the pressure build-up by molecular hydrogen, or through the concentration of hydrogen atoms and embrittlement phenomena at the crack tips and on grain boundaries. Both processes will have the same deleterious effect on the metal structural properties. Water chemistry, corrosion effects and hydrogen sources Hydrogen from cathodic partial corrosion reaction

Figure 4: Schematic diagram of hydrogen diffusion and blister formation.

Figure 5: Typical hydrogen induced cracks (source: MTI Atlas of Corrosion and Related Materials Failures – electronic ed.)

Hydrogen blistering or cracking is controlled by minimizing corrosion and is normally not a problem in neutral or alkaline environments and with high-quality steels that have low impurity and inclusion levels. Nevertheless, also under the primary water chemistry conditions of the reactor coolant system (RCS) of PWRs, with a typical pH T of approx. 6.9 to 7.4 at the operating temperature of ca. 300 o C (corresponding to a room temperature pH of around 10), the primary cathodic corrosion reaction will be:

- → H + OH -

H 2 O + e

because of the low proton concentration compared with that of water. Even for very low corrosion rates of the stainless steel cladding (e.g. 0.1 to 1 micron/yr) this will result in significant quantities of corrosion-generated hydrogen atoms that will evolve and may enter into the base metal of the RPV (> 10 24 - 10 25 atoms/yr). In this respect, Tomlinson 13 has shown that in oxygen-free, high-temperature water more than 90% of the hydrogen generated in the cathodic corrosion reaction is indeed absorbed by the steel. The austenitic stainless steel cladding is sometimes considered to prevent hydrogen diffusion and potential hydrogen-induced cracking problems in the pressure vessels. This, to our knowledge, has never been proven experimentally in an adequate way 14 and, at most, the cladding probably has only a “delaying” effect in transferring the nascent hydrogen to the cladding/base metal boundary, and further into the RPV steel matrix. The presence of flaws in this matrix (cf. “hydrogen flakes”) represents ideal, irreversible sinks (traps) for the hydrogen injected into the metal from the cathodic corrosion reaction.

Hydrogen from coolant radiolysis

In addition to the corrosion-generated hydrogen, there is also the issue of hydrogen radicals being formed as a result of the radiolysis of water and the reactions of H 2 with the radiolysis products 15 (e.g. OH ⋅ + H2 → H ⋅ + H2O); hydrogen is used in the RCS to suppress radiolytic oxygen and hydrogen peroxide formation. More details of all these effects are discussed elsewhere 16 .

Earlier international observations

Finally, there are earlier observations of hydrogen-induced blister cracking that have been reported in nuclear structural materials 17 , and in the past there has been considerable debate about the issue. A very old, specific, example of failures attributed to hydrogen occurred in retaining rings used to connect inlet assemblies to the reactor process tubes in a Hanford, water-cooled production reactor. Failures occurred in carbon steels and Type 420 stainless steel. The reported hydrogen sources were the fabrication process, hydrogen generated during corrosion of the ring by the process water, and from galvanic coupling. 18 In view of all of the above, the “trapping” of cathodically-generated hydrogen (due to primary water corrosion reactions), or from radiolytic hydrogen, inside existing “hydrogen flakes” is not improbable. Moreover, the (original) flakes may act as stress raisers, which will enhance the diffusion of the hydrogen to the stressed areas in the metal, possibly also resulting in hydrogen stress cracking (HSC). Also, the additional effect of irradiation is still largely unknown. Traditionally, it has been assumed that hydrogen significantly affects the fracture properties of pressure vessel steel in both the unirradiated and irraditated states at hydrogen contents above 2 ppm. Some studies have measured the hydrogen content of the cladding and found it to be 3-4 ppm after prolonged irradiation in PWR water. 19 This can also be assumed to be the equilibrium content at the cladding/base material interface. Although hydrogen diffuses quickly in the RPV steel at high temperatures, the presence of efficient hydrogen traps, such as the “hydrogen flakes”, poses a severe threat. The effect of irradiation and the importance of hydrogen in some observed low fracture toughness values clearly requires further research. Aging risks and crack growth

More details will be discussed during the presentation.

CONCLUSIONS

In June-July 2012 “thousands” of fissures were discovered in the Reactor Pressure Vessel (RPV) of the Belgian nuclear reactor Doel 3 during an ultrasonic inspection (UST). In September 2012 similar defects, but fewer in number, were also found in the reactor Tihange 2. Both RPV forgings were produced by the same fabricator which also delivered some 10 vessels to US nuclear plants. “Hydrogen flakes” originating from the processing of the original RPV ingots were identified as the root cause of the problem. After an initial series of investigations, the reactors were authorized to re-operate, until a number of anomalous embrittlement results were found in irradiation experiments on similar materials. After the stop of the reactors, new UST inspections

in 2014 indicated the presence of a total of 13,047 “hydrogen flaws” in Doel 3 and 3,149 in Tihange 2. However, some three years after the first detection of these “thousands of hydrogen flaws” in the RPV shells, and after new investigations, both reactors received authorisation to restart in December 2015. Since then, the affected reactors have been plagued by a number of scrams and unforeseen shutdowns. In view of all this, the potential problem of (hydrogen-related) crack-growth in the RPVs and the related longer-term aging problems of the reactors are still imminent and deserve further attention; probably more than ever. At this point in time, for example, it remains unclear if the cracks found in the Belgian NPPs Doel 3 and Tihange 2 are "only" manufacturing artifacts, or if there is also an "operational component" contributing to the current problems and operational risks; i.e. whether the cracks are still progressing and whether there are other phenomena, e.g. similar to 'hydrogen blistering' or hydrogen-induced cracking (HIC) processes, contributing to the problem. Additional hydrogen might indeed come from the cathodic corrosion reactions occurring on the primary water side of the reactor pressure vessel, or from radiolytic hydrogen of primary water decomposition. During operation, there is a permanent flux of (corrosion-originating or radiolytic) atomic hydrogen – although the flux might be small – and this hydrogen could easily become trapped into the voids that are present or are created at inclusions in the wall of the RPV. An eventual pressure build-up in the flakes will result in growing cracks and other materials degradation phenomena. It is to recognize that it is not just Doel 3 and Tihange 2 in Belgium that could be affected. 20 The RPVs were fabricated by the now bankrupt RDM (Rotterdamsche Droogdok Maatschappij, Netherlands), which also manufactured RPVs for at least 20 other reactors that are operating in seven countries around the world, including some 10 in the United States. Also, more recently, similar phenomena as in the Belgian reactors have been detected in the Swiss reactor Beznau 1 and, to a lesser extent, in Beznau 2. These RPVs were fabricated by a different manufacturer, which demonstrates that other factors, like steel supplier, cladding process and final assembling, may also have played an important role in the development of the observed damage. This being said, it should be recognized that a pressure vessel with a density of flaw “indications”, as were found in 2012 in both Belgian RPVs, would not have been accepted at the time of fabrication.

If some of the hypotheses discussed above were to be proven to be true, there might be a huge impact on currently operating PWRs world-wide.

REFERENCES

1 C. Peachey. Cracks found at Doel 3 . Nuclear Engineering International, p. 10, October 2012.

2 N.N. (Bel-V). Flaw indications in the RPVs of Doel 3 and Tihange 2 – Safety Evaluation Report R-SER- 13-001-0-e-0 (Internal report), 34 pp. (January 2013).

3 N.N. (AIB-Vinçotte). Synthesis Report Doel 165 (Internal report), 10 pp. (January 2013).

4 N.N. (Electrabel). Safety Case Report: Doel 3 – Reactor Pressure Vessel Assessment (December 2012).

5 N.N. (Electrabel). Safety Case Report: Tihange 2 – Reactor Pressure Vessel Assessment (December 2012). 6 N.N. Doel 3 and Tihange 2 reactor pressure vessels – Final Evaluation Report. FANC (Federal Agency Nuclear Control, Belgium), 33 pp. (May 2013). 7 N.N. Defects in the reactor pressure vessels of Doel 3 and Tihange 2 . The Greens / European Free Alliance (European Parliament), 41 pp. (March 2014). 8 E. van Walle. The detection of hydrogen flakes in the Belgian Doel3/Tihange2 reactor pressure vessels – Overview of technical developments to support restart justification. NENE 2013, Bled, Slovenia, September 2013. 9 N.N. Flaw indications in the reactor pressure vessels of Doel 3 and Tihange 2 – Progress Report 2014. FANC (Federal Agency Nuclear Control, Belgium), 10 pp. (December 2014). 10 R.J. Fruehan. A review of hydrogen flaking and its prevention . Iron & steelmaker, Vol. 24, Nr. 8, p. 61 (1997). 12 R. Boonen & J. Peirs. Critical reflections about the integrity of the reactor vessels of the Doel 3 and Tihange 2 nuclear power plants. Unpublished Report, 21 pp. (September 2016). 13 L. Tomlinson. Mechanism of corrosion of carbon and low alloy ferritic steels by high temperature water. Corrosion, Vol. 37, Nr. 10, p. 591 (1981). 14 R.E. Mazel, V.G. Grinenko & T.P. Kuznetsova. 1980. Influence of cladding on the resistance to hydrogen embrittlement of the metal of a nuclear reactor vessel. Teploenergetika, Vol. 27, Nr. 10, p. 29 (1980). 15 D.D. Macdonald & M. Urquidi-Macdonald. The Electrochemistry of Nuclear Reactor Coolant Circuits , Encyclopedia of Electrochemistry, Vol. 5 Electrochemical Engineering, Eds. D.D. Macdonald and P. Schmuki, Wiley-VCH Verlag, pp. 665-720 (2007). 16 W.F. Bogaerts. J.H. Zheng. A.S. Jovanovic & D.D. Macdonald. Hydrogen-induced Damage in PWR Reactor Pressure Vessels. Corrosion 2015, Research in Progress Symposium, NACE (2015). 18 R.E. Westerman. An analysis of the Truarc Ring Failure Problem , HW-71631, November 8, 1961; cited in: S.H. Bush & R.L. Dillon , ‘Stress Corrosion in Nuclear Systems’ , Proc. Int. Conf. on Stress Corrosion Cracking and Hydrogen Embrittlement of Iron Base Alloys (NACE-5), p. 61, June 1973. 19 J. Koutsky & K. Splichal. Hydrogen and radiation embrittlement of CrMoV and CrNiMoV ferritic RPV steels. Int. J. of Pressure Vessel and Piping, Vol. 24, p. 13 (1986). 20 N.N. Activities in WENRA countries following the Recommendations regarding Flaw Indications found in Belgian Reactors. WENRA (Western European Nuclear Regulators Association), 27 pp. (December 2014). 11 R. Boonen. Private Communication, June 2016. 17 T.K.G. Namboodhiri. Trans. Indian Inst. Metals, Vol. 37, p. 764 (1984).

Nuclear Engineering and Design 129 (1991)331-339

331

North-Holland

Behavior of pressure-vessel steels exposed to hydrogen from corrosion and environment

H. Pi r che r Thyssen StahlAktiengesellschaft, Duisbur~Germany

Received7 August 1989, revisedversion5 March

1990

Examples reflecting materials performance under service conditions are used to illustrate the potential deterioration of pressure-vessel steels exposed to hydrogen under varying corrosion and stress conditions. The fundamental reactions of hydrogen attack at elevated temperatures and pressures, of hydrogen-inducedcracking, and of various types of hydrogen-in- duced stress corrosion cracking are discussed. Steel selection possibilities, constructional and processing parameters, corrosive-mediuminfluences, and protectivemeasures are presented.

1. InWoduction

cracking (HIC) or hydrogen-induced stress corrosion cracking (HSCC).

Under service conditions, some metallic materials tend to pick up hydrogen as a result of reactions with environmental media. For steels with a bcc structure, i.e. with a ferritic, ferritic-pearlitic, bainitic or marten- sitic structure, this may occasionally lead unexpectedly rapidly to a deteriorat ion of the service propert ies and occasionally to a catastrophical failure. Well-known ex- amples are reflected in such problems as are caused by hydrogen at tack under pressure at temperatures above 200 °C in high-pressure reactors for ammonia produc- tion or coal hydrogenation. But, spectacular failures have also occurred as a result of hydrogen-induced corrosion during the exploitation or refining of wet, sour hydrocarbons. Hydrogen-induced corrosion of steel always takes place in two steps [1]. At first, atomic hydrogen is picked up by the steel from the surrounding media (aqueous solutions or gases). Fol lowing the concentra- tion gradient, the absorbed hydrogen then diffuses into the interior of the steel where, as the second step, hydrogen-induced cracking occurs at critical spots. Wi th respect to cracking by hydrogen from corrosion, a dif- ferentiation can be made between chemical and physical reactions. An example reflecting the first reaction is the reduct ion of cementite in C -Mn steels causing methane to originate from hydrogen at tack at elevated pressure and service temperatures [2,3]. The result of physical reactions are fissures stemming from hydrogen-induced

2. Hydrogen attack at e l evated pressure and tempera- tures above 200 ° C

An outstanding example showing the effect of hy- drogen at elevated pressure and temperature on steel is known from the Nobel-prize paper presented by C. Bosch in 1932 [2,4]. In the process of the first catalyt ic high-pressure hydrogenat ion of coal, the format ion of methane caused a plain-carbon steel fitting to undergo decarburizat ion followed by a slackening of the micro- structure and finally by brittle fracture. The causes of this type of hydrogen corrosion which is also called hydrogen at tack (HA) have been studied thoroughly, and extensive experience has been gained from analy- sing the damage sustained, from testing the materials used and from laboratory trials performed. They have been published by the American Petroleum Inst i tute in the form of the so-called Nelson diagrams; excerpts of the 1983 edition are shown in fig. 1 [3]. The higher the service temperatures and the hydrogen part ial pressures, the greater the percentage of such allowing elements as chromium, mo l ybdenum and occas iona l ly also vanadium and tungsten in steels resistant to hydrogen at tack at elevated service temperatures and pressures. Unl ike cementite, carbides containing these alloying elements present the resistance required to prevent

0029 - 5493 / 91 / $03 . 50 © 1991 - El sevi er Sc i ence Pub l i she r s B.V. (No r t h -Ho l l a n d )

332

tt. Pircher / ttydrogen corrosion of pressure-uessel steels

800

i

/

r j o .c_ 600 0 0 ~, 400 E 0

~. , . , .~.~. ~ . ~ 1,25Cr-O,SMoSteer

6 OCr-O5MoSteel

1 - .

-.J

30Cr-0S.o,eel

"...,,,~Cr- 0.5MoSteet ~

~

J ~20Cr-OSMoStee't

,

t

. . . . .

~

. . . . .

{ '

I

I

i

I

I

I

I

'

/

~

~

1 . 25Cr -O. SMo or 1,0Cr -O, SMo S t e e l

+

"

I

1~SMo Steer

~ - -

Corbon ~tee[

-

~ J.

200

I

T

T

T T T T T T TT 200 345 621 896 207 ~B3 758

50

100

150

Hydrogen portial pressure in ba r Fig. 1. Operating limits for steels in hydrogen service.API-Nelson diagram [3].

surface decarburization (dashed lines) and internal de- carburization with cracking due to the formation of methane (solid curves).

tr'

3. Hydrogen solubility and di ffusion

iO I

Fig. 2 shows the concentrations of hydrogen (defined by the Sievert solubility constant S) which are soluble in the lattice of ferritic ferrous materials [11]. From the figures stated it is possible to infer the hydrogen con- tents cn for different equilibrium pressures by way of using the Sievert equation c . =S The solubility of hydrogen in steel decreases sharply as the temperature drops. Shut down of high-pressure reactors after a prolonged period of operation at elevated temperatures tends to result in hydrogen oversaturation of the materials, which has to be removed by way of effusion. The latter is governed by the laws of diffusion which make it necessary to take into account also decreasing diffusion coefficients with decreasing tem- perature. If excess hydrogen remains in the material after cooling down to the ambient temperature, delayed cracking occurs, similar to flaking as a result of metal- lurgical hydrogen introduced during steelmaking (fig. 3).

E r~

lda

iO o

•~-- t n

r -

U

G) 8

t6"

o

o c/~ I 0-a

=

ii+i+i+i

!i i6-•

16 l°

Io-'

| 0 0 0

~0 0

100 Tempe r o t u r e in °C

Fig. 2. Solubility constant and diffusion coefficient of hydro- gen in low-alloysteels [11].

4. Hydrogen- induced cracking

liquid media is favored by promotors. Fig. 4 displays an absorber [6] used to dry wet sour gas at the exploitation site prior to feeding it into the pipeline for conveyance

As shown in fig. 3, a very similar pattern is obtained whenever the pick-up of large hydrogen quantities from

333

H. Pircher/Hydrogen corrosionofpressure-vesselsteels

16

2

E E 10

I i i S%NQCI.H2Sriemd, pH3

E ~ e E o_ .c 6

Fig. 3. Flakes in a hot-rolled 25 mm plate.

E " 2

o.o3~'o,,," Synma t ~ mm wa t * r . N2 r i ns * ~ . l ~ |

|

-

o . o~g~ ,

I , / ,

o:

,

, ° I

, o

o

,o

~

3o

,o

so

6o

",o

=

90

,oo

after Riecke Time in h Fig. 5. Permeation rates of hydrogen in the pipefine steel grade X 60 for various solutions [11].

to the cent ral gas refining equipment . The spots marked on the shell are ul t rasonical ly detected flaws at t r ibuta- ble to hydrogen- induced cracking. At omi c hydrogen is present as a result of the cathodic react ion of an electro- lytic corrosion whose absorpt ion by the steel is favored by H2S as a promotor . The amoun t of hydrogen picked up by the steel surface under given cor ros ion condi t ions can be de- termined by means of permea t ion measurements . This is i l lustrated in fig. 5 where a C -Mn steel is used as an example under free cor rosion in three di f ferent aqueous solutions. The effect exer ted by the promo t or s - - H2S in the present case - - and by the pH value of the medium becomes evident . The decreasing hydrogen per- meat ion rate wi th t ime is due to the forma t ion of protect ive layers. Wi thout external tensi le stress existing, the great amount s of hydrogen absorbed by means of promo t or s tend to form bl isters near the surface and cracks inside the material at non-metal l ic inclusions or microst ruc- tures having a critical shape and hardness. In plates, cracking preferably occurs in planes paral lel to the rol led surface (fig. 6). Structures suscept ible to hydrogen- induced cracking are made up of di f ferent types and shapes of non- metal l ic inclusions, in par t icular manganese sulfides, and of hard banded structures in segregat ion zones.

Fig. 4. Absorber with HIC in service [6].

334

H. Pireher / Hydrogen corrosion of pressure-vessel steels

At climatic temperatures and under mechanical load with no plastic deformation, such hydrogen absorption from aqueous solutions under service conditions as tend to endanger the pressure-vessel material will generally occur only in the presence of promotors. Hydrogen absorption from gases within the aforementioned range of temperatures presupposes a very clean and active steel surface which does not exist in technical construc- tional parts. Catalytic effects such as those caused by palladium, for instance, can mostly be excluded under technical service conditions. Clean and active surfaces are likely to be brought about by creeping under alternating mechanical loads or in notches. Under plastic deformation with a low strain rate, hydrogen is picked up from all media able to offer hydrogen. Here, no promotors are required. The method mostly used to determine the HSCC resistance of a steel is to expose stressed tensile speci- mens in a suitable electrolyte solution at ambient tem- perature, for instance in accordance with NACE stan- dard TM-01-77 [10]. Because of the relatively small amount of equipment required, HSCC testing of steels and welds is often done on bend test specimens as well. Fig. 7 shows test results obtained from weldable structural and tube steels. Plotted on the x axis is the yield strength of the materials and on the y axis the threshold stress for HSCC after 720 hours of testing [5]. The steels can be divided into two strength categories which present a different behavior with respect to the HSCC resistance. For steels with a yield strength of up to 650 N / mm 2, the critical stress increases as the yield strength rises. Under the stringent test conditions, threshold stresses from 40 to 90% Rel have been de- termined. For high-strength steels with a yield strength above 650 N / mm 2, an inverse trend was noted. Here, the threshold stress for HSCC decreases as the yield strength rises, i.e. the steels become more susceptible to HSCC. In connection with the HSCC susceptibility of welds fig. 8 provides test results obtained from three-point bend test specimens. The specimens were taken from the top layer of welded joints and from single-pass welds in different types of steel plate. The test results are plotted against the maximum hardness of the heat- affected zone. Differentiation is made between speci- mens that present cracking in the heat-affected zone during the 400 h test period, and those that passed the corrosion test with no HSCC or with cracking of the 6. Promotor-enhanced HSCC at ambient temperature

Fig. 6. Hydrogen-induced cracking (HIC); (a) Blisters, (b) Stepwise cracks [11].

Metallurgical measures to be taken to improve the resistance to HIC are aimed at lowering the contents of sulphur and oxygen in steel, and at controlling the inclusion shape. In this respect, paramount importance is attached to the ladle treatment using calcium, a metallurgical process developed in the late sixties [7,8]. Further steps relate to the avoidance of hard banded structures by way of reducing the carbon and manganese contents or by preventing excessive carbon segregation during the cooling from the final austenizing of the steel. With a view of furnishing evidence of an increased resistance of steel to HIC, exposure tests are generally carried out in an H2S-containing acid medium. Subse- quent to an exposure which mostly lasts for 96 h, the test specimens are thoroughly checked for blistering and internal cracking. The method most frequently used is described in the NACE standard TM-02-84 [9]. Apart from the absorption of hydrogen by the steel, hydrogen-induced stress corrosion cracking (HSCC) also presupposes the existence of tensile stresses of a critical magnitude which depends upon the amount of hydro- gen absorbed. Cracking is preferably oriented in a direc- tion perpendicular to the direction of principal stress. Stress-induced HSCC exists whenever the hazard is exclusively the result of exceeded threshold tensile stress, in contrast to strain-induced HSCC where plastic defor- mation within critical strain rates is required. 5. Hydrogen-inducedstress corrosion cracking