210 likes | 313 Views
Progress on PWR Lower Head Failure Predictive Models _________________. V. Koundy 1 , F. Fichot 1 , H.-G. Willschuetz 2 , E. Altstadt 2 , L. Nicolas 3 , JS. Lamy 4 , L. Flandi 4 1 . IRSN – 2 . FZD – 3 . CEA – 4 . EDF (SARNET WP10-2 WORKING GROUP) .
E N D
Progress on PWR Lower Head Failure Predictive Models _________________ V. Koundy1, F. Fichot1, H.-G. Willschuetz2, E. Altstadt2, L. Nicolas3, JS. Lamy4, L. Flandi4 1. IRSN – 2. FZD – 3. CEA – 4. EDF (SARNET WP10-2 WORKING GROUP)
Background (exp. and num. studies on LHF) Benchmark calculations Presentation of the different models used Comparison of the different results Conclusions Presentationoutline
Background SA assessments Late phase of a postulated SA Definition of appropriate accident mitigation strategies Loads A good understanding of the mechanical behaviour of the RPV lower head • Experimental programmes : LHF/OLHF(SNL), FOREVER(RIT),… • Modelling investigations (2D simplified, 2D FE and 3D FE)- 2D calculations (Benchmark) - 3D calculations by Cast3m (complementary results)
Calculation benchmark“OLHF-1 experiment” - 1/5 scale LH experiment- 1/2 scale wall thickness - Thermocouples along 4 meridians: ● 3D calculations ● 2D calculations (0°longitude) - Uniform heating (corium gradually relocates to the LH) - Experimental findings (failure time/location…) - Severe necking (from 76mm to 1mm) - T° as a function of time at measurement points - Transient internal pressure
Pressure, dead weight T° on the I/O surfaces (given by the thermocouples) T° in the wall (thermal solution) Temperature loading Presentation of the models used1. The FZD FE model (in ANSYS) Axisymmetric FE model432 four-node elements ● Plasticity ● Creep/Damage modelling(Fortran routines) 1. 2. a creep DB (D : J. Lemaitre) ● Failure criterion : D=1 Material DB REVISA & OLHF Programmes
● Creep/Damage modelling • : Norton-Bailey (Sandia) • 2. D : Kachanov – by post-evaluation (properties : RUPTHER) • ● Failure criterion : eq limor D=1 Presentation of the models used2. The CEA FE model (in CAST3M) Axisymm. FE model576 eight-node elts. Pressure, dead weight T° on the I/O surfaces (given by the thermocouples) T° profile in the wall (linear) Temperature loading
● Creep/Damage modelling 1. (Norton-Bailey) 2. D : Kachanov formula ● Failure criterion : D=1 (Similar to that used by CEA) Presentation of the models used3. The EDF FE model (in Code_Aster) Axisymm. FE model1152 eight-node elts -Very fine mesh-Elt. n° doubled OLHF & Rupther Programmes Pressure, Temperature T° Thermocouple readingsT° in the wall (thermal model)
Ro r ● Theory of shells of revolution under symmetric loading ● Large displacements, large strains : 0 /2 100 intervals Thickness 11 calculation points ● Creep/Damage modelling 1. : Norton-Bailey (Sandia) 2. D : Kachanov formula (Rupther) ● Failure criterion : D=1 (Like in the CEA model) Presentation of the models used4. The IRSN simplified model (in Astec) • The ruptured vessels remain axisymmetric (like in LHF and OLHF) • Hemispherical LH Ovoid (egg-shape) Pressure, Temperature(T° in the wall : thermal model)
OLHF-1 Benchmark Comparisonof different 2D calculation results
Longitude 180° 0° Nodal temperatures given by SNL 0° Uz=0 12.3 MPa 90° Ur=0 OLHF-1 Benchmark Boundary and loading conditions Hemispherical part –Meridian section at a longitude of 0° – Uz=0 andUr=0 Constant pressure – Nodal temperatures (thermocouples)
OLHF-1 Benchmark Calculated – experimentally determined results () All the calculations : damage criterion (except Cast3m) Cast3m stopped prematurely “necking criterion” ( the damage was negligible at the predicted time ) Certain results of Cast3m will not be compared to the other results
17.7cm 16.9cm 14.6cm (TEST) 70-75° (failure locations) 9.5cm OLHF-1 Benchmark Comparison the elongation as a function of time • Similar fail. locations (latitudes 70-75°) Test: latitude of 75° • trup texp(discrepancy value =3%) • Uz : discrepancy value =35% (fast Uz) • Code_Aster, Ansys~experimental curve
OLHF-1 Benchmark Wall thickness profile along a meridian line (at t=trup) 0° • CEA, FZD-initial wall thickness profile (thicker at the top)-“curves”similar to “exp. curve”- min. thicknesses over-estimated • Aster, Astec- « curves » ~ exp. Curve (polar angles up to 70°)-not parabolic behaviour 90°
0° 90° 160 OLHF-1 Benchmark Comparison of different calculated Vm (int. & ext. surfaces at the pole) as a function of time • FE codes (good agreement) • Simplified model- stress values underestimated - quite good results after 160 min
Benchmark conclusion Calculation Benchmark 2D models : - able to correctly predict the exp. failure time and failure location. - the global mechanical behaviour of the OLHF-1 experiment was well represented (by all the models). - not able to give information relating to crack propagation and breach size ( 3D models are necessary for these issues). An IRSN/CEA joint programme Complementary study of crack propagation and crack size using a 3D model of Cast3M
Crack opening and propagation3D calculations using Cast3m ●Constitutive law Unified coupled damage viscoplastic model (Lemaitre-Chaboche) ● Failure criterion : D=1 or Eps=100% 11488 CU20 (20-node elements) ●Crack propagation modelling -Depressurisation (perfect gas)-Damaged elements removed from the mesh Variation of initial wall thickness Thermocouples : 4 meridian lines Top ~ 20°C and Linear axial T° gradient in the cylinder Small cylindrical extension Pressure, dead weight, T° loading
17.6 cm2 14 cm2 Uz=15.7 cm Uz=14.6 cm 3D calculations Model : Interpretation of the OLHF/LHF experiments Yes OLHF-1 correctly interpreted No NB: LHF material (more sulpher) behaved in a brittle manner / OLHF material Exp. failure section > Calculated failure section Unsatisfactory interpretation of the LHF experiments The 3D model validated by the OLHF experiments is unsuitable for the LHF experiments Need of a failure criterion which can take into account the variability in material behaviour
Conclusions • 2D Calculations are able to interpret the global mechanical behaviour of lower head failure experiments. Failure time and failure locationcan be correctly estimated. • For crack opening and propagation, 3D calculations are necessary, but the failure criterion must take into account the variability in material behaviour. The development of such a criterion is one of the objectives of the current joint research programme between IRSN, CEA and INSA Lyon.
0° 40° 90° OLHF1 Benchmark Comparison of different results
OLHF1 Benchmark Comparison of different results
OLHF1 Benchmark Comparison of different results Maximum local measurement of the exp. Value : ~ 140%