PSA 2017- September 2017 IRSN Severe accident risk assessment for Nuclear * Power Plants * Enhancing nuclear safety An original approach to derive seismic fragility curves - Application to a PWR main steam line Nuclear Safety Division Severe Accidents Department Nadia RAHNI Marine MARCILHAC Maria LANCIERI Georges NAHAS Julien CLEMENT Lionel VIVAN Yves GUIGUENO Emmanuel RAIMOND Thomas CHARTIER Thierry YALAMAS
Contents 1. Introduction 2. Ground motion 3. Mechanical models 4. Mechanical failure criteria 5. Uncertainties and sampling 6. Seismic response of the steam line 7. Fragility curve evaluation 8. Conclusion 2/28
1. INTRODUCTION 3/28
1. Introduction For several years, IRSN (acting as Technical Safety Organization for the French Nuclear Safety Authority) has been developing L1 and L2 PSAs for the French PWRs IRSN is performing research to extend the scope of its existing L2 PSA to internal or external hazards Concept of confinement PSA is applied: the objective is to calculate conditional failure for the confinement function depending on the external hazards intensity IRSN has developed a methodology to determine the seismic fragility curve of a component (steam line) supported by a structure (containment building), by means of a detailed approach involving a large number of numerical calculations -> 4 steps : 1/ develop suite of seismic time histories representing variation of ground motion spectra 2/ build fast-running numerical models of the supporting structure and the component and define failure criteria 3/ propagate uncertainties and compute the structure and component mechanical responses 4/ compare the component responses to failure criteria and derive fragility curves 4/28
2. GROUND MOTION 5/28
2. Ground motion Acceleration time histories (accelerograms) are considered as inputs for nonlinear response analysis issued from probabilistic seismic hazard analysis assessment (PSHA) of a specific NPP site PSHA has been performed for six return periods from 1,000 to 10,000,000 years logarithmically spaced with a specific calculation at 20,000 years For each return period : uniform hazard spectra (UHS) is assessed to the average and to 5 fractiles: 5 %, 16 %, 50 %, 84 % and 95 % For all the 36 computed UHS, 360 three-components accelerograms have been generated by matching real acceleration time histories to uniform hazard spectra Input motion is applied at the base of the supporting structure modelling Peak Ground Acceleration (PGA) has been chosen to characterize seismic ground motion level 6/28
2. Ground motion Uniform Hazard Spectra distributions 0,15 g < PGA < 2,25 g Real signals Matched signals 7/28
3. MECANICAL MODELS 8/28
3. Containment building Double-wall structure : reinforced prestressed concrete wall (inner wall) and reinforced concrete wall (outer wall) inner wall : represented with 4 sticks (Timoshenko beam models), each stick represents a quarter of the containment outer wall and internal structures : represented with simple sticks located at the structure s center of gravity Mechanical characteristics of the stick models have been identified and validated from the respective finite elements 3D model These simplified models allow fast calculations (essential for performing several hundred runs) Containment behavior is considered as linear elastic at first approximation (valid in the small deformation range) Numerical simulations are performed with CAST3M (finite elements software ) 9/28
3. Containment building Inner wall 3D modelling Outer wall 3D modelling Internal structures 3D modelling Inner wall stick modelling Outer wall stick modelling Internal structures stick modelling 10/28
3. Steam line Line section modelled : between the steam generator (inside the containment) and the stop downstream from isolation valve (outside the containment) modelled with right pipe, elbow and beam elements taking in consideration the pipe line and several valves, supporting devices and stops at different elevations A previous analysis showed that : 1/ the maximum stress is located in the containment penetration area development of an additional local model for the penetration, considering the nonlinear behavior of the steel material 2/ the vertical stop is identified as a critical point results obtained from the main pipe model at the vertical stop are specifically analyzed Isolation valve Vertical stop Containment penetration 11/28
3. Steam line Concrete Supporting devices Penetration sleeve Vertical stop Containment penetration Shroud 12/28
3. Soil structure interaction Introduced with help of a 6x6 stiffness matrix applied at the center of the lower face of the raft Based on the use of impedance functions (characterizing the response of the foundation to a unitary signal in the frequency domain) Impedances are computed assuming that the footing is circular, rigid, massless and superficial Given to geological context of the considered site, the footing soil comprises an alluvial layer of thickness H resting on an infinite half-space of marls 13/28
4. MECANICAL FAILURE CRITERIA 14/28
4. Mecanical failure criteria Possible failure modes are failures of support or rupture of piping due to excessive stress (the low number of cycles due to the seismic loading does not justify a fatigue study) Step 1/ Verification of the steam line integrity for each time and each node of the line, a linear calculation is performed equivalent stress is calculated according to the French nuclear construction code requirements and compared to the allowable stress (387.6 Mpa) if the allowable stress is not reached, it is considered that the steam pipe line is able to resist the seismic load Step 2/ In case of exceedance of the allowable stress a new calculation is performed taking into account the nonlinear behavior of the steel material the total plastic deformation is compared to allowable strain of steel for the pipe and the containment penetration (between 17 % and 24 % at 300 C) In addition : effort calculated at the model s node corresponding to the vertical stop is compared to the stop design effort (775 kn) 15/28
5. UNCERTAINTIES AND SAMPLING 16/28
5. Uncertainties and sampling 15 uncertain parameters are identified Parameter Interval Description H (m) [24-27] Upper layer thickness Soil r 1 (kg/m 3 ) [1800-2100] Upper layer density Containment building Steam line r 2 (kg/m 3 ) [1900-2100] Half-space density E IC (MPa) [27700-45556] ξ RPC (%) [4-6] ξ RC (%) [6-8] e 1, e 2, e 3, e 4, e 5, e 6 (mm) [29.8-38.3], [33.3-42.8], [34.1-43.9], [33.3-42.8], [53.4-68.6], [34.1-43.9] Modulus of elasticity - Inner containment Reinforced prestressed concrete damping Reinforced concrete damping Pipe thicknesses ξ SL (%) [1-4] Steam line damping Containment penetration R (mm) [30-40] Flange s radius Earthquake N acclerogram [1-360] Accelerogram number 17/28
5. Uncertainties and sampling Confidence intervals and distributions are chosen depending on the parameter knowledge (measurements, specification notes, bibliography etc.) The 360 accelerograms have the same probability of occurrence : uncertainty on earthquake occurrence is taken in consideration during the convolution of the fragility curve with seismic hazard curves, to derive the annual failure frequency of the confinement (not addressed in this presentation) An experimental design is build using the LHS technique : sampling is done considering the space of uncertain parameters and the total number of earthquakes a 360x15 matrix is generated each row allows to define a specific calculation with the CAST3M software 18/28
6. SEISMIC RESPONSE 19/28
To help protect your privacy, PowerPoint has blocked automatic download of this picture. To help protect your privacy, PowerPoint has blocked automatic download of this picture. 6. Seismic response Alluvial layer H Marls Acceleration time histories ξ ssss Soil-structure interaction Containment building displacement Steam line response (pipe, vertical stop) Containment penetration response 20/28
6. Seismic response Maximal equivalent stress in the pipe line obtained for each of the 360 linear calculations (not necessarily obtained at the same time and the same location) show that : for PGA greater than 0.7 g and for some runs : allowable stress is exceeded New non-linear calculations using the pipe model and the containment penetration model are then performed for cases leading to exceedance of the threshold total plastic deformations < 3 % : steam pipe line and containment penetration integrity is not challenged Maximal effort obtained at the vertical stop show that the threshold is reached for a set of runs This response is then chosen to build the fragility curve of the vertical stop 21/28
7. FRAGILITY CURVE EVALUATION 22/28
7. Fragility curve evaluation Fragility curve P f a = P X F tt a) probability that for a given PGA a, the maximal effort at the vertical stop F exceeds a given threshold tt Methodology based on linear regression developed in the context of non-destructive tests that aim at detecting flaws in a structure by computing curves of probability of detection as a function of flaw size Two steps Data linearization F = g(f): find an appropriate transformation g providing the finest linear regression of F a, F (a) = β 0 + β 1 a + ε Exceedance probability computation P f a Probabilistic/statistical methods 23/28
7. Fragility curve evaluation Data linearization F (a) = β 0 + β 1 a + ε with ε the error model Find β 0,β 1 and σ ε 2 g β 0 + β 1a Initial data Transformed data 24/28
7. Fragility curve evaluation Exceedance probability computation P f a 1. Compute the law of β 0 and β 1 conditionally to σ ε 2 2. Compute a sample of (β 0, β 1, σ ε 2 ) 3. Compute a sample of (P f a ) -> fragility curve The hypothesis on the error model ε has a very low influence on the results ϵ normally distributed ϵ estimated by Kernel smoothing 25/28
8. CONCLUSION 26/28
8. Conclusion IRSN has developed a methodology to derive seismic fragility curves of a component supported by a structure With the implemented methodology, the fragility curves are directly built from a large number of simulations in a sufficiently short time to make them usable in practice By comparison to the safety factor method which relies on expert judgments and fragility parameters, this study is expected to provide method for realistic seismic risk assessment In order to increase quality of results, future challenges will concern the taking into account of local geological conditions (site effects) and non-linear behavior of the containment Otherwise, a specific modelling of the vertical stop should be carried out to ensure relevance of the analysis 27/18
Thank you for your attention (nadia.rahni@irsn.fr) 28/28