Critical Reflections about Doel3 & Tihange2

Critical reflections about the integrity of the reactor vessels of the Doel 3 and Tihange 2 nuclear power plants.

Authors: dr.ir.ing. Boonen René email: rene.boonen@kuleuven.be dr.ir. Jan Peirs email: janpeirs1@gmail.com

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 2

Preface

For an engineer who has followed the news about the reactors of Doel 3 and Tihange 2, it was quite amazing and against common engineering sense that these two reactors are put in service again after three years of research, whereby more than 13000 flaws in the reactor vessel shells have been re- ported. This forms the cause to search for relevant information. A number of reports concerning the reactor vessels of Doel 3 and Tihange 2 are present at the FANC website http://www.fanc.fgov. be/nl/page/dossier-pressure-vessel-doel-3-tihange-2/1488.aspx?LG=2 . Unfortu- nately, these reports are difficult to read, they are tedious, the information is very fragmented and spread out over a large number of different reports. In order to understand how such an investi- gation is carried out, it is recommended to read a report about another case of Structural Integrity Associates Inc. [ 1 ] . This 82 page report deals with a crack of a weld inside a reactor vessel, it ex- plains clearly the course of the investigation, provides the data of the materials and the load cycles, cites where necessary the approach and the data from the "ASME Boiler and Vessel Code", describes the calculations and presents the results in such a way that technical skilled people outside the nu- clear sector can understand these calculations. This is not possible with the material provided in these reports on the FANC website. To do calculations, data from the report of Structural Integrity Associates Inc. [ 1 ] are used, as these data lack in the reports provided by FANC. The findings will be illuminated in this note. First, the construction of the pressurized water reactor (PWR) will be illuminated such as it is presented in the reports of Electrabel. Then, some principles of fracture mechanics will be explained. Next, the most important issues from the Electrabel reports will be illuminated. This will be done at hand of the reports itself, the reader must have these reports at his disposal to follow the explanations, such that a biased viewpoint will be avoided. Then, the critical attention points will be formulated. The consulted sources will be cited by their internet addresses in the bibliography where possible, such that the reader is capable to find them quickly. The authors are totally independent from any organisation or political party and want to keep it that way. The authors believe that nuclear energy will be indispensable in a low–carbon economy. It will not be possible to fully replace the current production of electricity by wind turbines and solar cells [ 2 ] . These energy sources consume too much space in our highly populated areas and varies too strong with the weather conditions. Nuclear energy can be safely used on the condition that normal engineering rules will be followed with reactors which follow the state of the art and that exploitants and government accept that every technical installation has a finite lifetime and needs to be replaced by new ones which comply with the actual state of the art. An accident in the primary circuit of a nuclear reactor, even when the consequences remain inside the nuclear plant itself, will make nuclear energy impossible in the future.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 3

Contents:

p3: Construction of a nuclear reactor vessel. p4: Main points of criticism on the Electrabel safety cases p6: Identification of the flaws in the reactor vessel. p12: Hydrogen as cause of metal cracking. p17: Fracture mechanics. p18: Treatment of flaws by Electrabel. p20: Chronology of the events leading to the nuclear code case N-848. p23: Minor issues. p25: References. p28: Communication with FANC.

1 Construction of a nuclear reactor vessel.

Figure 1 presents the construction of a nuclear reactor vessel [ 21 ] . In the Electrabel–report [ 14 ] p17- 21, the construction of a nuclear reactor vessel is also described. The reactor consists of a cylindrical steel vessel with a diameter of 4 m, a wall thickness of 200 mm and a height of 13m. It is a PWR–reactor ("Pressurized Water Reactor"). The water in the reactor is pressurized by an external pressurizer at 155bar such that the water does not boil in the reactor. This primary water is heated by the nuclear core and circulated by pumps through the steam generators with a volume flow of 21190 m 3 / h in each loop, which produces the steam for the turbines which generate the electricity. The electrical power amounts 1020MW. The primary water enters the reac- tor vessel at a temperature of 282 o C, flows between the core barrel and the vessel wall downwards and subsequently back upwards through the core and leaves the reactor vessel at a temperature of 325 o C. The reactor vessel is designed to withstand a pressure of 171 bar and is tested at a pressure of 215 bar. Figure 2 (figure 3.5 in the report [ 14 ] p20) presents the construction of the reactor vessel of Doel 3.

The vessel has been welded together from different forged parts. The reactor has started up in 1982 and is about 34 years in use.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 4

Figure 1: Principle of a nuclear reactor [ 21 ] .

2 Main points of criticism on the Electrabel safety cases

1. Hydrogen as cause of the flaws. The normal volume H 2 necessary to form 41 cracks in 1dm 3 steel at cracking pressure amounts between 818 ml and 1374 ml, which is 6 to 10 times more than the 131 ml H 2 which was dissolved in the steel. However, not all hydrogen takes part in crack formation. It is estimated that only 61ml H 2 is available for crack formation. Then 13 to 22 times 61ml H 2 is necessary to form the cracks. This would implicate that hydrogen flaking cannot be not the only cause of the flaws, but that another cause has resulted in flaking and / or that the flaws have been growing during time. This is discussed in more detail in section 4.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 5

Figure 2: Components of the Doel 3 reactor pressure vessel and vessel head [ 14 ] p20.

2. The absence of a consolidated theory for material with a high density of cracks. The fracture mechanics theory mainly deals with the behaviour of a single crack under stress with- out interaction of neighbouring cracks. In recent research, the interaction between two cracks have been published in several articles, and multiple crack situations are still in the research phase. There are no precedent cases published wherein such high crack densities in construc- tions occur. Electrabel has proposed grouping rules which led to the ASME BPVC nuclear code case N-848. In this code case, flaws in close proximity to each other are grouped in rectangu- lar boxes. The circle or ellipse with the largest dimension fitting in this box is then treated as a single ”equivalent flaw”, which then will be evaluated according to the ASME BPVC criteria. No proof has been found if the ”equivalent flaw” is the worst case flaw in all possible flaw combinations in the vessel wall. It makes the nuclear reactor vessels of Doel3 and Tihange2 an experimental case for a non–consolidated fracture mechanics theory. This is discussed in more detail in section 6.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 6

3 Identification of the flaws in the reactor vessel

In 2012, flaws have been detected during an ultrasonic inspection. These flaws are located mainly in the lower core shell of the Doel3 reactor vessel. The number of flaws amounts 7205 ( [ 14 ] p32). The distribution in position with respect to the inner wall is presented in figure 3 (figure 4.7 in [ 14 ] p34) and the distribution in size is presented in figure 4 (figure 4.21 in [ 14 ] p55).

Figure 3: Distribution of the flaws with respect to the inner wall distance [ 14 ] p34.

Figure 5 (figure 4.6 on [ 14 ] p33) shows the distribution of the flaws in the lower core shell. The left side shows an actual cross-section measurement, the right side shows the flaws within a sector of 20 o . Figure 6 (figure 4.16 in [ 14 ] p47) presents an image of the flaws in the lower core shell for different sectors. The sectors between 200 o [ 14 ] p47 and 220 o exhibit a high flaw density. Figure 7 presents the distribution of the flaws around the circumference of the lower core shell, measured in 2012 and in 2014. The mean flaw density between the first 100 mm from the inner reactor surface amounts 2.1 / dm 3 and the maximum flaw density is 25.8 / dm 3 ( [ 15 ] p7). The mean flaw distance is 20.4 mm, the maximum distance is 565 mm ( [ 15 ] p7). The smallest flaw distance was not mentioned in these reports. In 2014, a new ultrasonic inspection has been performed. During this inspection, a higher number of flaws has been found, i.e. 11607 instead of 7205 in the lower inner shell. ( [ 16 ] p27). Also the reported flaws appear to be larger, the mean size found in the 2012 inspection was 9.6 × 7.6 mm, in 2014 16.0 × 12.7 mm and the largest flaws in 2012 67.9 × 38.4 mm, in 2014 179.0 × 72.3 mm ( [ 16 ] p29). In figure 8 (4.9 in [ 16 ] p28), the new distribution of the flaw distance to the inner vessel wall is presented and in figure 9 (4.11 in [ 16 ] p30) the distribution of the flaw size. In the reports

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 7

Figure 4: Distribution of the flaw size with respect to the inner wall distance [ 14 ] p55.

is mentioned that this increase in number and size of the flaws is due to the increased resolution of the measurement equipment used in 2014 compared to the equipment used in 2012 and not to a physical increase of the flaws in the reactor vessel. ( [ 17 ] p11). Based on the new flaw data, the mean flaw density within the first 100 mm wall thickness should amount 3.4 flaws / dm 3 and the maximum flaw density 41.6 flaws / dm 3 ( [ 15 ] p7).

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 8

Figure 5: Picture of the flaws in the lower core shell, on the right a local cross section, on the left a collection over a sector of 20 o [ 14 ] p33.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 9

Figure 6: Picture of the flaw distribution in the lower core shell in different sectors [ 14 ] p47.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 10

Figure 7: Distribution of the flaws over the Doel3 lower shell circumference, measured in 2012 (upper part) and in 2014 (lower part). [ 20 ] .

Figure 8: Comparison of the measurements of the depth distribution of the flaws with respect to the inner wall distance in 2012 and 2014 [ 16 ] p28.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 11

Figure 9: Comparison of the measurements of the flaw size distribution in 2012 and 2014 [ 16 ] p30.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 12

4 Hydrogen as cause of metal cracking.

In this section, it will be demonstrated that the hypothesis of Electrabel that hydrogen segregation which occurred during the vessel production as the only cause of metal cracking cannot be correct. To demonstrate this, the amount of available hydrogen for cracking has to be estimated, then, the amount of hydrogen necessary to cause all the cracks has to be estimated. If the amount of hydrogen available for cracking is smaller than the amount of hydrogen necessary to cause all the cracks, the hypothesis that hydrogen flaking during the manufacturing process of the shell as the only cause cannot be correct.

Figure 10: Solubility of hydrogen in steel in terms of temperature and pressure [ 11 ]

The ingots of which the reactor shells have been manufactured were poured using a state-of-the-art vacuum technique, to remove the hydrogen from the molten steel. Before casting the ingot from which the lower core shell of the Doel3 reactor has been manufactured, the molten steel has been

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 13

sampled and the hydrogen content was measured [ 14 ] p43. After cooling down the ingot in a controlled manner, samples of the metal are taken and the hydrogen content was measured again. The amount of hydrogen present in the steel of the Doel3 lower core shell was for all measurements 1.5 ppm [ 14 ] p44. During cooling of the ingot, impurities and alloying elements segregates into so called ghost lines which are sensitive to hydrogen cracking [ 9 ] . As solidification progresses from the outside wall inwards, the internal part of the ingot will be enriched with impurities and alloying elements, while the outside part will be poor in alloying elements, due to the higher solubility of the alloying ele- ments in the liquid phase. During forging, the parts of the ingot with the high segregation areas have been removed by cutting off the bottom and the top from the ingot and in a next step the centre part of the ingot which also contain high segregation areas has been pierced out. From the resulting hollow cylinder, the reactor vessel shell has been forged. After forging and cooling until room temperature, the inner and the outer wall will be machined by cutting away about 40 mm at each side on a lathe. Hypothesis of Electrabel. The hypothesis of Electrabel, as presented in the paper [ 9 ] and also in their communication on page 28, is that when the segregation areas finally transform from the γ -phase to the α -phase during cooling of the ingot, they become supersaturated in hydrogen which will recombine to hydrogen gas H 2 at trapping sites in the metal, building up an internal pressure. The combination of this internal pressure and local stresses can lead to cracking during or shortly after fabrication of the shells. So, Electrabel concludes that the cracks were already present at the start-up of the reactor. Available amount of hydrogen in case of a uniform distribution over the shell. Figure 10 presents the solubility of hydrogen in iron, where the curve of 1 atm should be followed. The necessary hydrogen concentration on the curve in figure 10 is 5 ppm at 900 o C before the γ - α - transformation starts to be supersaturated in the α -phase. However, as the hydrogen concentration has been measured in the melt to be 1.5ppm and afterwards in the cooled metal also to be 1.5ppm, concentration of hydrogen did not occur as the metal has never become saturated, the level of hydrogen present in the metal remains at least 3 times below the saturation level. The saturation of hydrogen occurs on the curve in figure 10 at 400 o C at a hydrogen level of 1.5ppm. It is unrealistic to

Total dissolved available H 2 volume

concentration H 2 volume

for cracking (Nml / dm 3 )

(Nml / dm 3 )

(ppm)

threshold concentration uniform concentration

0.8 1.5

70

–

131

61

Table 1: Volume of normal cm 3 H

3 steel

2 in 1 dm

assume that each individual atom hydrogen diffuses towards the crack zones. Some of the hydrogen will escape to the atmosphere, another part will be trapped on the crystal boundaries and inclusions in the metal without initiating crack formation, and a part remains in solution. In order to have an estimate of the amount of hydrogen which does not take part to crack formation, the amount of

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 14

hydrogen is taken from the solubility curve presented in figure 10 where the curve of 1 atm crosses the abscissa, which corresponds to 0.8 ppm at 100 o C, which corresponds to 70ml H 2 / dm 3 . Table 1 summarizes the amounts of hydrogen in the steel. The last column of table 1 presents the amount of hydrogen involved in the flaw creation process.

Figure 11: Axisymmetric 2D-FEM model of a crack.

Figure 12: Von Mises stress distribution around a crack loaded by an internal pressure inside the crack from the center to the edge of the crack.

Necessary amount of hydrogen to cause the high density of cracks in the shell. In the next paragraphs, it will be demonstrated that there is not sufficient hydrogen dissolved in the metal to

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 15

20

15

5 crack ceiling displacement ( µ m) 10

0

0

2

4

6

8

10

radius from centre of crack (mm)

Figure 13: Displacement of the crack wall as result of an internal pressure of 3796 bar, from the center to the edge of the crack.

create such a high density of cracks present in the high crack zones of the Doel3 lower shell. In order to obtain a first estimate the amount of hydrogen needed to generate the flaws in the high flaw density material, it is assumed that 41 flaws of a mean diameter of 17mm are present in 1 dm 3 steel. The flaw is caused by accumulating hydrogen at a hydrogen trap (impurities, segregation zones) in the vessel wall material and at a certain pressure, the material cracks. During cracking of the material, the pressure in the crack drops until the material stress intensity factor (SIF) drops below the threshold value and the crack gets its final dimensions. For a crack of 17 mm diameter, the internal pressure inside the flaw equals for are circular shaped flake σ = π 2 K I c p π a , [ 4 ] in which K I c = 40 MPa / p m is the stress intensity factor at room temperature and a = 8.5 mm is the half diameter of the flaw. To estimate the volume of such a crack, a linear elastic finite element simulation has been carried out. Figure 11 presents the axisymmetric 2D-FEM model of a circular crack of 17mm diameter in the center of a block steel of 400mm diameter and 800mm high, such that the crack can be considered as if it is present in an infinite medium. Only the bottom is fixed, all the other boundaries are free. Inside the crack, the pressure is introduced which should remains directly after the cracking, in this case 3796 bar. The resulting stress distribution around the crack is shown in figure 12, where the stress concentration at the crack tip is visible. Figure 13 presents the resulting displacement of the crack upper boundary, from which the volume can be estimated. The volume of such a crack amounts 0.00525ml. The normal volume H 2 from 41 cracks at 3796 bar pressure amounts then 818ml, which is 13 times higher than the 61ml H 2 which

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 16

Size Number Content H 2 pressure H 2 Volume H 2 Total Volume (mm) – (ml) (bar) (Nml) (Nml) 10 2000 0.001373 4948.9 6.80 13590 15 4000 0.003829 4040.8 15.47 61887 20 2300 0.007970 3499.4 27.89 64145 25 1100 0.01400 3130.0 43.82 48202 30 800 0.02220 2857.3 63.42 50734 35 400 0.03291 2645.3 87.04 34818 40 200 0.04613 2474.5 114.14 22828 45 150 0.06218 2332.9 145.05 21758 50 100 0.08126 2213.2 179.85 17985 55 80 0.10362 2110.2 218.66 17493 60 50 0.12946 2020.4 261.56 13078 65 30 0.15903 1941.1 308.70 9261 Total 11210 – – – 375777

Table 2: Distribution of the cracks in the Doel3 lower shell, with the volumes and pressure of hydrogen to cause them.

was available for flaw generation. If the size distribution of the cracks is equal to the distribution over the complete lower shell, the result will be as follows. The lower shell contains 11607 indications ( [ 16 ] p27). The distribution to size in the Doel 3 lower shell has been reconstructed from the graph ( [ 16 ] p30) and is presented in table 2. The finite element analysis has been carried out for each different flaw diameter. The corresponding pressures and volumes of hydrogen necessary to form the flaws are tabulated in table 2. When it is assumed that the high density area has the same distribution as the whole lower shell presented in table 2, the volume of H 2 to cause the cracks will be 375777 11210 · 41 = 1374 ml / dm 3 steel, which is approximately 22 times the hydrogen available for flaw formation. This leads to the conclusion that there is not sufficient hydrogen dissolved in the steel to generate such high density of cracks. In the complete lower shell with a wall thickness of 200mm, a diameter of 4m and a height of 2.5m, 823 l H 2 is dissolved of which 383 l H 2 is available for cracking for the uniform case. This is slightly more than the 375 l H 2 necessary to cause all the cracks. Then this hydrogen has to diffuse several meters following the circumference of the reactor vessel to form flaws in the high density areas. It is more likely that a large part of the hydrogen diffuses out of the 200mm thick wall or being trapped in the metal than to travel a few meters to the high flaw density areas. In both calculations, the amount of dissolved hydrogen initially in the steel is far too small (13 to 22 times too small) to cause such high density of flakes. This leads to the conclusion that hydrogen flaking cannot be the only cause of the flaws, but that another cause has resulted in flaking or that the flaws have been growing during time.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 17

5 Fracture mechanics.

The phenomenon of crack propagation has been subject to extended research in different mechanical applications, particularly in reactor and aviation applications. A crack in a metal does not propagate without reason, it needs energy to propagate. This energy is supplied as elastic energy due to the mechanical stresses as result of the reactor’s operation. The growth of the crack consumes a part of this elastic energy. As cousins phenomena, a part of the elastic energy will be dissipated in plastic deformation, a small part in surface energy and a small part in kinetic energy of the moving metal, generating sound waves in the metal. A brittle material uses the majority of the available elastic energy for crack growth and a small part for plastic deformation. A ductile material is the opposite, the majority of the available elastic energy is used for plastic deformation and only a small part for crack growth. For this reason, crack growth is slow in ductile materials. As criterion for crack tip stresses, the stress intensity factor (SIF) is introduced. In the SIF, the crack geometry and the bulk stress is taken into account. This SIF will then be evaluated to the critical SIF, which is determined as a material property at hand of tensile and Sharpy impact tests. There is a lot of literature available about this subject. However, most of the literature looks very specialized, but there is also some good non-specialist literature [ 4 ] , [ 5 ] , [ 6 ] , [ 7 ] . Three types of loading modes can occur: 1. Crack opening mode, mode I: The material tension is perpendicular to the crack and pulls the crack further open. The stress intensity factor of SIF is K I = βσ p π a , wherein β is a geometry factor, σ the bulk tensile stress and a is the half length of the crack. The SIF will be evaluated to K I C , which is a material parameter which is a measure for ductility. It is determined by tensile and Sharpy impact tests. 2. Crack shear mode, mode II: The material tension is parallel to the crack and and shears the crack further open. The stress intensity factor of SIF is K I I = βτ p π a , wherein β is a geometry factor, τ the bulk shear stress and a is the half length of the crack. In principle, this SIF will be evaluated to K I I C , but in most cases, it is converted to mode I and evaluated to K I C . 3. Crack tearing mode, mode III: In this mode the opening is torn sidewards. This mode does not appear in this investigation. This theory is based on the research of one single crack in the material, i.e. there is no interac- tion with other cracks. Studies about multiple cracks in the material are very recent and is not consolidated in a profound theory. In most of these studies, two cracks with interaction are under investigation. Some studies point out that this aspect is not fully covered in the ASME Boiler and Vessel Code and that the effects of multiple cracks on the strength of the vessel is not sufficiently conservative estimated [ 8 ] ,p363-364, [ 3 ] .

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 18

6 Treatment of flaws by Electrabel

The absence of a consolidated theory about multiple cracks and interaction between cracks, and the absence of publications of other applications of such a theory on other cases such as aviation or other reactor vessels (chemical industry) is the main point of concern. Electrabel has conducted research to multiple cracks and has developed grouping rules for closely spaced flaws [ 18 ] ,p18. As far as the author could trace the method to group these flaws from their publications, one of the publications on which the grouping method Electrabel has developed is based on is the work of Kunio Hasegawa, Koichi Saito and Katsumasa Miyzaki [ 22 ] . A similar study has been carried out by Ali Abbaszadeh Bidokhti and Amir Reza Shahani which is published in the Latin American Journal of Solids and Structures [ 23 ] . Hasegawa et al. are searching for alignment rules to decide if cracks which are in close proximity to each other should be treated as non-aligned or as coplanar. The criteria for these alignment rules are defined in many fitness-for-service codes (AMSE, JSME, etc. . . ), however, these criteria are different in the different codes. Based on finite elements simulations and tensile tests on steel specimen with two parallel cracks in opening mode at different distances from each other, they propose new alignment rules for cracks for linear elastic fracture mechanics evaluation.

Figure 14: Illustration of flaw grouping [ 14 ] p74.

Based on the paper of Kunio Hasegawa, Koichi Saito and Katsumasa Miyzaki [ 22 ] , Electrabel de- veloped grouping rules for flaws which are published in two papers which are presented at two different conferences [ 24, 25 ] . This work has led to the proposal of an ASME code case N–848 "Alternative characterization rules for quasi-laminar flaws – Section XI, Division I" [ 26 ] , [ 27 ] . Using these newly developed rules, a part of the flaws will be grouped in rectangular boxes ( [ 14 ] p71-75) based on the proximity of the flaws to each other as presented in figure 14 (figure 4.37 in [ 14 ] p74). Then, the circle or ellipse with the largest dimension, as illustrated in figure 15 (figure 4.38 in [ 14 ] p75), will be taken as an ”equivalent flaw” which should be the worst case flaw and will then be evaluated as a single flaw using the ASME BPVC criteria. There is no proof found in the literature or in the reports available at the FANC website which

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 19

Figure 15: Equivalent flaw as result of the flaw grouping [ 14 ] p75.

Figure 16: Flaw content of a grouped flaw with 69 mm equivalent size [ 13 ] p16.

demonstrates that the "equivalent flaw" is really the worst case flaw, which could imply that some flaw groups are less conservatively evaluated than the individual flaws. For example, a large flaw together with some other flaws in its shadow could result in a box with the same dimensions as the large flaw, i.e. in such a case the effect of the other flaws in the box is not considered. Also branched flaws will not be considered, although it is an almost certainty that with such a high flaw density in

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 20

the vessel walls, branched flaws are present. An example of a grouped box is presented in figure 16 (figure in [ 13 ] p16). This box has as dimen- sion 69 mm and is composed of 7 flaws of which 4 flaws with a dimension of 11 mm, 1 flaw of 13 mm, 1 flaw of 14 mm and 1 flaw of 42 mm. To what extend this boxed flaw behaves as a single flaw of 69 mm is not described in the FANC reports. As stated before, current research concerns the interaction of two or a few flaws [ 8 ] ,p361, but never in such flaw density as reported in de Electrabel reports about the reactor vessels. The Oak Ridge National Laboratory (ORNL) performed a detailed technical review of the 2015 Electrabel Safety Cases prepared for the Belgium reactor pressure vessels at Doel3 and Tihange2 [ 18 ] . The ORNL carried out an independent quantitative assessment of the entire flaw popula- tion in the two Belgian reactors according to the ASME Boiler and Pressure Vessel Code, Section XI, Appendix G, "Fracture Toughness Criteria for Protection Against Failure", New York (1992 and 2004) using ORNL tools, methodologies, and the ASME Code Case N-848, "Alternative Character- ization Rules for Quasi-Laminar Flaws". The ORNL also pointed to an internal Electrabel report CNT-KCD / 4NT / 20374 / 000 / 01, dd. 30 / 07 / 2014, wherein proximity rules for flaws are been inves- tigated [ 18 ] p18. In the investigation of the ORNL, the majority of the flaws are treated as single flaws according to the ASME BPVC criteria. The other flaws were grouped according to the code case N–848, the grouped flaws were treated as a single flaw with as dimension the largest circular dimension of the resulting box diagonal. For the flaw groups which were found non-compliant to the ASME BPVC acceptance criteria after screening, the circle is replaced by an ellipse fitting into the 3D-boxes (see figure 15, the ellipse in dashed line) and screened again. [ 18 ] p23. The grouped flaws which do not comply the ASME BPVC acceptance criteria are listed in tables 5.2, 5.3, 5.4, 5.5 and 5.6 in the ORNL report [ 18 ] p39-45. After the new screening of the elliptical equivalent flaws, the flaw groups which were found non- compliant to the ASME BPVC criteria, the flaws previously in a group are no longer treated as a group, but are modelled individually in a finite element multi-flaw model that should account for mechanical interactions among closely spaced flaws. Particularly for the latter group, the results are questionable as in these cases, there is no con- solidated fracture mechanics theory available for multiple closely spaced cracks with interaction between these cracks and no precedent cases have been published. It makes the nuclear reactor vessels of Doel3 and Tihange 2 an experimental case for a non- consolidated fracture mechanics theory.

7 Chronology of the events leading to the nuclear code case N-848

As the nuclear code case N–848 is of prime importance to the analysis of the flaws in the Doel 3 and Tihange 2 nuclear reactor vessels, the chronology of the events which led to this code case has been reconstructed as well as possible by the authors using internet information. In this section, the time sequence is represented with the internet addresses which lead to technical documents, which

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 21

should allow the reader to find the appropriate information rapidly.

1. July 20, 2014: Publication of the conference paper: LACROIX, V., DULIEU, P., and COUPLET, D., "Alternative Characterization Rules for Quasi-Laminar Flaws", PVP2014-28200, ASME PVP Conference, Anaheim, CA, July, 2014." [ 24 ] , http://proceedings.asmedigitalcollection.asme.org/proceeding.aspx?articleid= 1937821 This paper is based on the paper of Kunio Hasegawa, Koichi Saito and Katsumasa Miyzaki [ 22 ] , and has led to the proposal of an ASME code case N–848 "Alternative characterization rules for quasi-laminar flaws – Section XI, Division I" [ 26 ] . 2. June 23, 2015: Approval of the code case N848 "Alternative Characterization Rules for Quazi-Laminar Flaws" https://cstools.asme.org/csconnect/CodeCaseForm.cfm?Action=View&CaseNumber= 2420&NoToolbar=yes and click on "View the Code Case File" for downloading the code case. or https://www.asme.org/wwwasmeorg/media/ResourceFiles/AboutASME/WhoWeAre/ BPVCResources/BPVC-CC-NC.pdf 3. July 1, 2015: Publication of the the code case N–848 "Alternative Characterization Rules for Quazi-Laminar Flaws – Section XI, Division I" http://standards.globalspec.com/std/9971057/asme-bpvc-case-n-848 4. July 19, 2015: Publication of the conference paper: V. LACROIX, P. DULIEU and A. S. BOGEART, "Alternative Characterization Rules for Quasi-Laminar Flaws Based on 3D X-FEM Calculations" ASME PVP Conference, July 2015 [ 25 ] . http://proceedings.asmedigitalcollection.asme.org/proceeding.aspx?articleid= 2471890 5. November 2015: Publication of the ORNL report ORNL / TM-2015 / 59349 "ORNL Evaluation of Electrabel Safety Cases for Doel 3 / Tihange 2: Final Report (R2)" http://info.ornl.gov/sites/publications/files/Pub59349.pdf or www.fanc.fgov.be/GED/00000000/4000/4030.pdf In this report, the ORNL applies the code case N–848 to recalculate the entire flaw population in the Doel 3 and Tihange 2 reactors according to the ASME Boiler and Pressure Vessel Code, Section XI, Appendix G, "Fracture Toughness Criteria for Protection Against Failure", New York (1992 and 2004).

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 22

6. November 12, 2015: FANC publishes its final report [ 19 ] and authorizes the Doel 3 and Tihange 2 reactor units to resume operation until they reach the age of 40 years ( [ 19 ] p6). www.fanc.fgov.be/GED/00000000/4000/4027.pdf

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 23

8 Minor issues

Figure 17: Effect of the flaws on the temperature distribution of the vessel wall.

1. Temperature gradient over the vessel wall in the flaw zone. The reactor vessel is ther- mally isolated such that the temperature gradient over the vessel wall amounts about 10 o C. However, the zones with a high density of flaws obstructs the thermal flow and a higher temperature gradient occurs at these areas, which leads to increased shear stresses (approx. 2.5 MPa / o C) at the crack tips. Figure 17 illustrates that effect.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 24

Figure 18: Effect of unequal distributed flaws on the roundness of the reactor vessel.

2. Boundary conditions for flaw simulation in shell segment. As far as the authors could understand the modelling of the flaws, a small segment from the reactor vessel shell has been selected for finite element simulation. The boundary conditions are the stresses resulting from the internal pressure in the reactor and the top of the segment can move parallel with the bottom. The latter boundary condition holds in the case the rest of the shell is solid, however due to the high flaw density, this can be an underestimation of the resulting stresses. [ 18 ] p77-81. Also the fact that only a segment is selected can lead to an underestimation of the resulting stresses. The hoop stress in the vessel wall is only pure tensile in the case that the vessel is perfectly cylindrical and the material is homogeneous. Due to the flaws, the material isn’t homogeneous any more, which has the same effect as unroundness, and additional bending stresses are introduced. This effect is demonstrated in figure 18. 3. Vibration levels are unknown. 4. Opinion of Westinghouse. The reactor is designed by Westinghouse, however the author has not found any opinion of Westinghouse about the implications of the flaws found in the reactor vessels in the reports available at the FANC website.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 25

References

[ 1 ] "Edwin I. Hatch Nuclear Plant - Unit 2 Reactor Pressure Vessel2N2G Recirculation Inlet Nozzle- To-Safe End Weld Full Structural Weld Overlay Crack Growth Calculations and Stress Analysis Spring 2011 Outage", Structural Integrety Associates Inc. http://www.ewp.rpi.edu/hartford/~wuy8/EP/other/References/ Ref5ML111880660.pdf [ 2 ] Hans-Werner Sinn, "The Green Paradox", MIT Press, 2012, ISBN: 9780262016681 http://www.hanswernersinn.de/en/topics/GreenParadox [ 3 ] R. Daud1, M.S. Abdul Majid1, M. Afendi1, A.K. Ariffin, S. Abdullah, " INVESTIGATION OF ELAS- TIC STRESS SHIELDING DAMAGE INTERACTION BASED ON FITNESS FOR SERVICE (FFS) CODES ", International Conference on Mechanical Engineering Research (ICMER2013), 1-3 July 2013, Paper ID: P286 https://afendirojan.files.wordpress.com/2010/04/p286.pdf [ 4 ] Broek, David, "Elementary engineering fracture mechanics", Martinus Nijhoff Publishers, 1986, ISBN 90-247-2580-1. [ 5 ] P.J.G. Schreurs,"Fracture Mechanics", course notes 4A780, TUEindhoven, 2012 www.mate.tue.nl/~piet/edu/frm/pdf/frmsyl1213.pdf [ 6 ] David Roylance, "Introduction to Fracture Mechanics", Department of Materials Science and Engineering, Massachusetts Institute of Technology, 2001. http://ocw.mit.edu/courses/materials-science-and-engineering/3-11- [ 8 ] Ruslizam Daud, Ahmad Kamal Ariffin, Shahrum Abdullah and Al Emran Ismail, "Interacting Cracks Analysis Using Finite Element Method", chapter 13 in "Applied Fracture Mechanics", edited by Alexander Belov, ISBN 978-953-51-0897-9, 392 pages, Publisher: InTech, Chapters published December 12, 2012 under CC BY 3.0 license http://www.intechopen.com/books/applied-fracture-mechanics [ 9 ] Evy De Bruycker, Séverine De Vroey, Staf Huysmans and Jacqueline Stubbe, "Phenomenology of Hydrogen Flaking in Nuclear Reactor Pressure Vessels", DOI 10.3139 / 120.110580, Materials Testing, 56 (2014) 6, Carl Hanser Verlag GmbH & Co. KG, ISSN 0025-5300 [ 10 ] Zhiye Chen, "Quality of as-cast ingots with extreme large shapes", Doctoral Thesis, Fakutät für Georessourcen und Materialtechnik der Rheinisch-Westfälische Technische Hochschule Aachen, December 9, 2014 mechanics-of-materials-fall-1999/modules/frac.pdf [ 7 ] M. Patricio, Robert M.M. Mattheij, "Crack Propagation Analysis" www.win.tue.nl/analysis/reports/rana07-23.pdf

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 26

[ 11 ] E.E. Fletcher and A.R. Elsea, "The effects of high-pressure, high-temperature hydrogen on steel", DMIC Report 202, March 26, 1964. [ 12 ] WENDRA Report: " Activities in WENRA countries following the Recommendation regarding flaw indications found in Belgian reactors" December 2014 www.wenra.org/media/filer_public/2014/12/26/flaws_in_rpv_feedback_2014- 12-19.pdf [ 13 ] "REPORT ON INDEPENDENT ANALYSIS AND ADVICE REGARDING THE SAFETY CASE Doel 3: Reactor Pressure Vessel Assessment" www.fanc.fgov.be/GED/00000000/4000/3397.pdf [ 14 ] "SAFETY CASE REPORT: Doel 3: Reactor Pressure Vessel Assessment", 5 / 12 / 2012. www.fanc.fgov.be/GED/00000000/4000/3390.pdf [ 15 ] "SAFETY CASE REPORT: ADDENDUM: Doel 3: Reactor Pressure Vessel Assessment", 5 / 12 / 2012. www.fanc.fgov.be/GED/00000000/4000/3434.pdf [ 16 ] "SAFETY CASE 2015: Doel 3: Reactor Pressure Vessel Assessment" www.fanc.fgov.be/GED/00000000/4000/4023.pdf [ 17 ] "REPORT ON INDEPENDENT ANALYSIS AND ADVICE REGARDING THE SAFETY CASE 2015: Doel 3: Reactor Pressure Vessel Assessment" www.fanc.fgov.be/GED/00000000/4000/4025.pdf [ 18 ] "ORNL Evaluation of Electrabel Safety Cases for Doel 3 / Tihange 2: Final Report (R1)", www.fanc.fgov.be/GED/00000000/4000/4030.pdf [ 19 ] "Flaw indications in the reactor pressure vessels of Doel 3 and Tihange 2 Final Evaluation Report 2015", FANC-AFCN, 12-11-2015. www.fanc.fgov.be/GED/00000000/4000/4027.pdf [ 20 ] Eric 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 11, 2013 www.djs.si/proc/nene2013/pdf/NENE2013_106.pdf [ 21 ] MIT Nuclear Engineering Course, http://ocw.mit.edu/courses/nuclear-engineering/22-06-engineering-of- nuclear-systems-fall-2010/lectures-and-readings/MIT22_06F10_lec06a.pdf [ 22 ] Kunio Hasegawa, Koichi Saito and Katsumasa Miyzaki, "Alignment Rule for Non-Aligned Flaws for Fitness-for-Service Evaluations Based on LEFM", Journal of Pressure Vessel Technology, Vol 131 / 041403-1, August 2009.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 27

[ 23 ] Ali Abbaszadeh Bidokhti and Amir Reza Shahani, "Interaction Analysis of Non-aligned Cracks Using Extended Finite Element Method", Latin American Journal of Solids and Structures, Au- gust 3, 2015, http://dx.doi.org/10.1590/1679-78251664 [ 24 ] V. Lacroix, P. Dulieu and D. Coupet, "Alternative characterization rules for quasi-laminar flaws", Proc. of the ASME 2014 Pressure Vessels and Piping Conference, PVP2014, July 20-24,2014, Anaheim, California, USA, papernr. PVP2014-28200 [ 25 ] V. Lacroix, P. Dulieu and A. S. Bogeart, "Alternative Characterization Rules for Quasi-laminar Flaws Based on 3D X–FEM Calculations", Proc. of the ASME 2015 Pressure Vessels and Piping Conference, Boston, MA, July, 2015, PVP2015-45792 [ 26 ] Nuclear Code Case N-848, "Alternative characterization rules for quasi-laminar flaws", ASME BPVC.CC.NC.S2-2015, approved June 23, 2015, available on the "cstools.asme.org"–website, "Working Group on Flaw Evaluation (SGES) (SC XI)", Code Cases, N848 [ 27 ] HASEGAWA K., STRNADEL B., LACROIX V., "Introduction and definition of laminar flaws pro- vided by flaw evaluation code", Metal2015, Jun 3rd - 5th 2015, Brno, Czech Republic, EU

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 28

9 Communication with FANC.

The critical reflections have been discussed with the FANC experts. Prior to the meeting on Januri 26, 2017, FANC has sent a note which is presented directly below. During the meeting the hydrogen balance has been the main discussion point. Although saturation of hydrogen cannot occur during the cooling after production of the shell, some concentration of hydrogen in the crack sensitive areas should occur. After the meeting, an additional calculation has been carried out to obtain an estimate of the possible hydrogen concentration in the crack sensitive areas with the data available. It will appear that the hydrogen concentration gradient over the shell wall is insignificant. This calculation has been communicated with the FANC also. Finally, a set of questions has been sent to the FANC in order to get more accurate data, however, the FANC refuses a reply to these questions. As result, the FANC will not consider our critical reflections nor take any action to let investigate further the cause of the flakes in the reactor shells more profoundly. Below follows the communication with FANC: Answer from FANC to the critical reflections Doel3 and Tihange2 prior to the meeting of January 26, 2017. Issue 1: Hydrogen as cause of the flaws When discussing hydrogen flaking as the most likely cause of the indications detected in the Doel 3 and Tihange 2 reactor pressure vessels, the Safety Cases of Doel 3 and Tihange 2 (version dated 5 December 2012) mention that ” The measured hydrogen level in the liquid metal of around 1.5 ppm, could be above the threshold for hydrogen flaking, since the sulphur level is relatively low. AREVA recommends a conservative maximum allowed hydrogen content of 0.8 ppm. ” From this information, the authors of [ 1 ] deduce that an amount of 61 ml / dm3 of H 2 is set free during the cooling of the material. After some calculations, and by assuming that there is no sig- nificant diffusion of hydrogen coming from the rest of the base metal towards the high flaw density areas, the authors of [ 1 ] conclude that hydrogen flaking cannot be the only cause of the flaws, and that another cause has resulted in flaking or that the flaws have been growing during time. The reasoning used to conclude this is misleading, mainly because the two following assumptions are not correct: • an amount of 61 ml of H 2 is set free : to our understanding, the authors of [ 1 ] make a confusion between the concept of hydrogen solubility in steel and the concept of the hydrogen content threshold to avoid hydrogen flaking; • there is no significant diffusion of hydrogen coming from the rest of the base metal towards the high flaw density areas : hydrogen diffusion is on the contrary at the heart of the flaking formation.

More information is given below, especially to understand how hydrogen flakes develop.

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 29

As developed below, hydrogen flakes appear during the fabrication of heavy forgings, when these have a mean hydrogen content which exceeds some acceptable critical value or threshold. Different values can be found in the literature (mainly because it is dependent of the content of other chemical species, such as Sulphur), but typically, for modern clean steels, this threshold is considered to be about 0.8 ppm. This means that when the mean hydrogen content of the forging is below this threshold, hydrogen flakes will most probably not develop. On the other hand, when the hydrogen content is above this threshold, there is a high risk of hydrogen flaking in the forging. When such flakes develop, they generally appear in large quantities, even when the hydrogen content is only about 1.5 ppm. This is because the amount of hydrogen that will participate in this phenomenon is not equal to the difference between the mean hydrogen content of the ingot and the threshold of 0.8 ppm, as supposed by the authors of [ 1 ] , and the reason is explained below. The amount of hydrogen that can be contained in the steel is characterized by its solubility (which is a different concept from the threshold of 0.8 ppm). This solubility expresses the maximum amount of (atomic) hydrogen which can be dissolved into the material. As illustrated by Figure 1, the solubility of (atomic) hydrogen is much higher in liquid steel than in the solid metal. Furthermore, the solubility in the solid steel depends on the phase: the hydrogen solubility in the γ -phase is higher than in the α -phase. At room temperatures, the hydrogen solubility is much smaller than 0.8 ppm.

Figure 1: Temperature dependence of hydrogen solubility in iron

The solidification of the large castings is a complex metallurgical process and the resulting ingot is characterized by several particular regions, differentiated by their chemical composition. When liquid steel locally solidifies, for solubility reasons, alloying elements (solute) are rejected in the surrounding liquid, which become enriched in those elements. Thus, the first region to solidify in an

R.Boonen & J.Peirs

May 18, 2017

Integrity reactor vessels Doel 3 and Tihange 2

Page: 30

ingot is poor in solute (negative segregation) and the last region to solidify is rich in solute (positive segregation). Local areas of highly solute-rich liquid can also form small channels rising to the top of the ingot and are termed A-segregates of ghost lines after their solidification. During solidification of the casting, hydrogen segregates to the surrounding liquid areas due to the difference in hydrogen solubility between the liquid steel and the solid steel. However, after the solidification, when the in- got is maintained at the high temperature at which the forging will be performed, the steel structure is austenitic ( γ -phase). After forging, the forged component is cooled down but the resulting trans- formation from the γ -phase to the α -phase is not uniform in the whole volume of the component. Due to their higher enrichment in alloying elements, the segregated areas have a lower transfor- mation temperature, which means that there is a time-lag between the transformation to ferrite of the unsegregated areas and that of the segregated areas. Unsegregated regions are first trans- formed to ferrite and due to the lower hydrogen solubility in α -phase, the hydrogen in the (ferritic) unsegregated regions diffuses to the (austenitic) segregated regions where it accumulates. When the segregated regions finally transform to ferrite, they become supersaturated in hydrogen, which precipitates in molecular form at trapping sites such as inclusions, grain boundaries and microvoids, building up an internal pressure. The ghost lines that are the most enriched areas are the last to transform to α -phase. So, the precipitation of molecular hydrogen occurs preferentially at the trap- ping sites in the ghost lines and in particular at the MnS inclusions. It is generally accepted that the internal H 2 pressure at the trapping sites is not sufficient to cause micro-cracks (flakes). Additional stresses such as transformation stresses, local stress concentrations around defects (inclusions) and deformation stresses during forging are believed to promote flaking. Moreover, another parameter promoting the occurrence of flaking is a cracking-sensitive microstructure. With regard to that, the ghost lines have a high content in alloying elements (e.g., C, Mn, P, Mo) that are quenching elements and are therefore more sensitive to quenching, which promotes the formation of a martensitic struc- ture which, under untempered condition, has a brittle nature. For all those reasons, when flaking is present in a forging, the flakes are located preferentially in the ghost lines. Calculations to estimate the final dimension of the flakes and the hydrogen pressure needed to form these cracks are very difficult, since they should take into account dynamical effects due to the cracking itself as well as deformation stresses due to the forging process (as it is known that these latter play an important role in the hydrogen flaking mechanism). The results obtained by the simulations made in [ 1 ] and which did not take into account these effects are thus questionable. Finally, we can convince ourselves that hydrogen flakes can form in large quantities, even if the measured hydrogen level seems quite small. Indeed, the VB395, which has not been used in service, had a relatively similar measured hydrogen level and destructive testing showed that it contained large quantities of hydrogen flakes having similar dimensions as the ones lying in the Doel 3 and Tihange 2 reactor pressure vessels (RPV). Issue 2: The absence of a consolidated theory for material with a high density of cracks It is known that the driving force acting on a given crack (stress intensity factor) can be significantly affected by the presence of one or more cracks in the close neighbourhood. Depending on the rel- ative position and orientation of the neighbouring cracks, this interaction effect can either increase or decrease the stress intensity factor. When assessing the fracture strength of structures affected by multiple cracks, the classical procedure used by the fitness- for-service Codes for avoiding the

R.Boonen & J.Peirs

May 18, 2017