1. Irradiation Shift – Questions and Answers

2. Q1 – The need to estimate irradiation shift
Reactor Pressure Vessel (RPV)

3. Start of life toughness
Asdg afdgasdge afgteasrd artea srdg dfvcxbgsjs dzffg srthy Why do we need to estimate irradiation shift?
Quite simply it is because irradiation makes reactor pressure vessel materials less tough. All nuclear reactor components (Slide 2) are made from materials that are very resistant to service conditions. However, they may degrade through life due to the influence of the operating temperature, the pressure and temperature cycles, and the effect of the water in the circuits.
In the case of the RPV (Slide 3a) one of the the most important materials properties for safety is toughness. RPVs are made from well-established materials that have high toughness (Slide 3b) before the RPV is put into service. RPV materials may be subject to the same, or similar, degradation mechanisms as other reactor components. However, (Slide 4) large areas of the RPV are also exposed to high levels of neutron irradiation. This causes very fine scale microstructural changes in the material (Slide 5), which can considerably affect its mechanical properties. The good start of life properties gradually change with time and the nice start of life RPV may be much less tough at end of life (Slide 6).
The damage caused by irradiation is progressive (Slide 7); it does not saturate. The mechanisms of damage are complex and can depend on a large number of environmental and materials factors. There can be substantial differences in the rates at which toughness degrades between types of RPV and even between RPVs of the same design.
In some cases
Before they enter service, RPV materials have high toughness (resistance to fracture)

4. Neutron irradiation
Core
RPV
Shielding
Irradiation fluence (dose) is a measure of total exposure of the material to neutron irradiation
Irradiation flux (dose rate) is a measure of the rate of exposure of the material to neutron irradiation

5. Effect of neutron irradiation (simulation)
Dislocation loop
Solute atom cluster
Vacancy cluster (vacancy-solute atom complex)
Box size 29x29x29nm
Vacancies white; Interstitials black
Cu red; Mn green; Ni blue
With thanks to Stéphanie Jumel, EDF, France

6. Effect of irradiation and ageing
Irradiation gradually degrades the toughness of RPV steels (operation at temperature may have a similar but smaller effect, even without irradiation)

7. Irradiation damage Progressive; does not saturate
Complex, not fully understood
Irradiation damage may be affected by:
Environmental factors
Irradiation fluence (dose); flux (dose rate); time
Irradiation temperature
Material factors
composition: Cu, Ni, Mn, Si, P, C, N …
strength; product form (plate, forging, weld); microstructure
Can vary substantially between types of RPV, between RPVs of the same type and within individual RPVs

8. Variability of irradiation damage
Data are from surveillance and test reactor irradiations for a wide range of materials and irradiation environments
Charpy shift provides a measure of irradiation damage

9. Fracture assessment ΔT = Irradiation Shift
Fracture toughness (MPa Ö
m) vs. temperature ( o C)
250
ΔT = Irradiation Shift
End of life KIC
Ductile
200
Start of life KIC
KI (SIF) transient due to cooling
150
Toughness ΔT
100
Margin 50
Brittle
-200
-100
100
200
300
Temperature

10. High toughness shift reduces the margin
Fracture toughness (MPa Ö
m) vs. temperature ( o C)
250 ΔT
ΔT = Irradiation Shift
Toughness
200
Start of life
End of life
150
KI transient ΔT
100
Margin 50
Margin
-200
-100
100
200
300
Temperature

11. Q2 - Experimental approaches for irradiation shift determination
Ideally material would be cut from prolongations of the most important components of the RPV and machined into fracture toughness specimens which are irradiated in the RPV surveillance capsules during its service
Core
RPV
Surveillance capsule
Shielding

12. In practice use Charpy testing
Historical reasons
Fracture toughness testing was not developed when surveillance testing started
Surveillance specimens are loaded at start of life
Practical reasons
Charpy tests are simpler and require a smaller volume of material (surveillance capsule space is limited)
There are reasonably good correlations between fracture toughness and Charpy shift

13. Charpy test method
Rough fracture surfaces show that high levels of energy has been absorbed; the specimen was tough
Fractured specimen

14. Charpy test results Energy, J ΔT41J oC Upper shelf (ductile)
 – unirradiated
 - irradiated
Upper shelf (ductile)
ΔT41J

15. Q3 - Differences between shift measurement methods for WWER and other PWRs
The differences in methods are small
For WWER reactors the Russian PNAE G code requires that the Charpy transition reference energy is varied according to material strength
In US-based codes the reference energy level is fixed at 41J
However the reference energy level has little effect on shift in most cases
There are also small differences in Charpy testing standards
(There are also differences in methods of estimating start of life fracture toughness)

16. Q4 – Correlation between Charpy and fracture toughness shift
Charpy and toughness shifts are closely correlated, for all RPV steels:
ΔTFracture toughness ≈ ΔTCharpy
However the correlations are scattered and there may be some deviations from the 1:1 relationship

17. Correlations for US RPV steels
US weld metals (left) show a 1:1 Charpy/toughness shift correlation, but in base materials (right), toughness shift is 16% higher than Charpy shift. Note that there is significant scatter (from Sokolov and Nanstad – Ref 42 in IAEA report)

18. Correlation for Russian steels
ΔT0 and ΔTF are well-correlated, however …

19. … other evidence (VVER-440 materials)
Some data indicate that toughness shift in VVER materials may be larger than Charpy shift (From Brumovský et al)

20. Interpretation of correlations
Charpy and fracture toughness tests are very different from each other, so a 1:1 shift correlation is not obvious
A difference in the tests is that Charpy tests measure the energy absorbed to fracture, including both brittle and ductile components, the toughness test measures the brittle component only
Irradiation can affect both components to different extents depending on the steel. It may be that the Russian indexing procedure (higher referencing energy) is better for compensating for this factor

21. Q5 - Reliable shift estimation
Regulations
Regulations
Require
ΔPredicted – ΔActual > x
+ Margin for uncertainties
Shift + margin for fracture assessment
Shift + margin for fracture assessment
13.5m
Require
ΔPredicted – ΔActual > x
(Unknown) error (bias) in prediction is
ΔPredicted – ΔActual
(Unknown) error (bias) in prediction is
ΔPredicted – ΔActual
+ Margin for uncertainties
Actual shift (ΔActual)
(we do not know this)
Actual shift (ΔActual)
(we do not know this)
ΔPredicted for important locations
ΔPredicted for important locations
Reliable shift and correlation models
Reliable shift and correlation models
Model justification
Model justification
Accurate values and uncertainties of materials and environmental variables at important locations
Accurate values and uncertainties of materials and environmental variables at important locations
NPC

22. Q6 – Test reactor data Are widely used Very versatile, can choose
Irradiation temperature, flux, fluence, spectrum
Test specimen type (including large KIC)
Controlled experiments, accurate data
Very useful in developing understanding and data to supplement surveillance data

23. Example of test reactor – HFR Petten

24. Lyra facility in HFR

25. Time/flux differences in test reactors
Flux and fluence are for neutron energy greater than 1MeV
75y
15y 3y
0.5y 1m 1w
Possible RPV fluence after 75 years
Times to achieve fluence values at a given flux
y(ears)
m(onths)
w(eeks)
Mark Kirk P034 IGRDM-15

26. Important considerations for test reactor irradiations
Acceleration of irradiation damage rate may affect the irradiation shift at a given fluence
High flux can delay precipitation damage in Cu-containing steels
Increasing flux may change irradiation damage mechanisms, for example producing unstable matrix defects (UMDs) (Nanstad et al 2008)
Late blooming phases (LBPs) may be produced in very high fluence, very long time, low flux irradiations
As a result irradiation shifts estimated using test reactor irradiations may be different from those in power reactors
Test reactor data should be used with great care, particularly in cases when they might give lower shifts than those in power reactors

27. Q7 – Available irradiation shift models for predictions beyond current data
Predictions for extended lifetimes are needed for economic as well as safety reasons
If adequate surveillance capsules remain in the RPV, there is no problem from the point of view of safety
Accurate irradiation shift models are required if it is not possible to do adequate surveillance testing and for economic reasons

28. Q8 - Model development methods
Earliest models were essentially empirical – statistical fits to data
As understanding developed, particularly in the 1980s and later, several mechanistically-based correlation (MBC) models were developed
More recently, thanks to further developments in understanding and increased computer power, several groups have worked to develop multi-scale multi-physics (MSMP or MSP) models

29. Examples of early empirical models
Cottrell (1957)
ΔT = A(Φ)0.5
US NRC Regulatory Guide 1.99 Revision 1 (1977)
ΔT = [ (Cu-0.08) (P – 0.008)] x (Φ/1019)0.5
(Cu-0.08)=0 when Cu <= 0.08; similarly for the P term
ΔT = shift; Φ = fluence
Varsik and Byrne (ASTM STP )
ΔT = [377.9 log(CR) ] x (Φ/3x1019) (for welds)
CR = chemistry ratio
= {[1.5Ni + Si + 0.5C + 0.5(Mn – 0.5)]/( Mo)} x Cu

30. MBC model example (Debarberis et al)
The left-hand plot shows how the model shows the contributions from three components of irradiation damage; the right-hand plot shows how the model may be fitted to WWER steel data

31. MBC model example (Nikolaev, Yu)
Improved residuals for the Nikolaev models: VVER-440 (left, ○) VVER-1000 (right,●), compared with the Russian Regulatory Guide (+, ○)

32. MBC model example (EONY)
Matrix feature (e.g. loops)
Copper-rich precipitates
Effect of dose rate
Effect of product form
Effective copper content
Eason, Odette, Nanstad and Yamamoto (EONY) model

33. MBC Model example, new WWER-1000 model
Margin
Total shift
For base: m = 0,8; AF = 1,45 оС
For weld: m = 0,8; AF = 1exp(2Ceq), оС
Thermal ageing shift
F is fluence, n/m2 (E>0.5MeV)
F0 = 1x1022; t is time
Irradiation shift
Effect of composition for weld

34. Development of more accurate models
Empirical and MBC models are fitted to data; extrapolation outside the database (or interpolation into sparse data regions) may be unreliable
Accuracy (and confidence) can be increased by developing greater understanding of irradiation damage mechanisms and refining the models
Confidence can also be increased by validation testing (eg of material from decommissioned RPVs)
But MBC models cannot be fully validated because unexpected effects may occur for combinations of conditions for which there are no data
In principle, MSP models can be founded on known physics, can be fully validated and can make accurate predictions for materials and irradiation environments for which no data exist
However, it may take many more years before a sufficiently well validated MSP model is available

36. Current benefits of MSP models
Provide valuable insights into damage mechanisms
Provide a framework for collaborative research between different organizations and technical disciplines
Can help identify critical issues and guide design of future experiments

37. Q9 Consolidated conclusions from Capture
Irradiation shift for WWER-440 steels
Damage has three or four components, potentially including P, Cu, P-Cu interaction and non-solute dependent defects
Fluence exponents may be different from ⅓, and differ between the damage components
The Russian reference model [Δ = 800(P Cu)F1/3] may not be conservative
At fluences beyond RPV end of life, total damage may start increasing at a high rate; can be explained by a P segregation model
Shift at a given dose increases as flux reduces; research reactor data may be non conservative
Analytical modelling can describe the observed dependencies and relationship between irradiation and re-irradiation response

38. Consolidated conclusions
WWER-1000 steels
Damage increases with Ni and Mn, and reduces with Si
Fluence dependency is unclear, the exponent may be as high as 1
The Russian reference model for Ni >1.5% (Δ = AF1/3, where A = 20 for weld, = 23 for base) may not be conservative
Shift increases as flux reduces
Thermal ageing can have a significant influence on shift

39. Consolidated conclusions
Fracture toughness behaviour
Toughness shift may be 20% greater than Charpy shift for both base and weld
At high levels of irradiation damage shape of toughness vs temperature transition curve may change (flatten)
Shift modelling
Improved mechanistically-based correlation models and physical models are being developed
There are limits to the data available for modelling (quantity, limited variability of materials, appropriate flux levels).
Accuracy of model inputs (fluence, composition) have been assessed

40. Q10 Open issues Functional form for shift model Amaev et al
Nikolaev Yu.
Russian regulatory models
Debarberis et al analytical model
Eason et al (EONY) (part of)
US Reg Guide 1.99 Rev 2

41. Open issues Accuracy of models (outside surveillance data range)
Mechanisms of Ni effects and Ni-Mn interaction
Mechanism of Si effect
Possibility of UMDs at high flux
Possibility LBPs at low flux, long time
Unknown unknowns
Convergence of models for different types of steel
Relationship between CV and KIC shifts
Effect of irradiation on KIC transition slope
Integration of SOL and Δ models and uncertainties
Future direction for shift model development
More data to develop and validate new models
Late Blooming Phase?
Miller et al (2005)