NRC ESSI SimulatorBoris Jeremic, Robert J. Budnitz, Annie M. Kammerer
University of California, Davis; Lawrence Berkeley National Laboratory; United States Nuclear Regulatory Commission
Earthquake Soil Structure Interaction Modeling: Benefits and Detriments
I Improved accuracy of Earthquake Soil Structure Interaction (ESSI) modeling andsimulation benefits understanding of seismic behavior of Nuclear Power Plants (NPPs)
I Reduction of modeling uncertainty! In other words, we will know with much more fidelitywhat is the actual seismic response of an NPP
I Benefit: reduction of seismic energy input due to energy dissipation of nonlinear materialI Benefit and Detriment: seismic motion frequency content change
Introduction to the NRC ESSI Simulator
A collaborative project between University of California, Davis (UCD), LawrenceBerkeley National Laboratory (LBNL), and United States Nuclear RegulatoryCommission (U.S. NRC) to develop a High Fidelity Modeling and Simulation Systemfor Time Domain, Nonlinear Finite Element Analysis of Earthquake Soil/Rock StructureInteraction (ESSI) response of NPPs.
NRC ESSI Simulator System Components
I The NRC ESSI Program: a 3D, nonlinear, time domain, parallel finite elementprogram specifically developed for High Fidelity modeling and simulation of EarthquakeSoil/Rock Structure Interaction (ESSI) problems for NPPs on NRC-ESSI-Computer.
I The NRC ESSI Computer: a distributed memory parallel computer, a cluster ofclusters with multiple performance processors and multiple performance networks.
I The NRC ESSI Notes: a hypertext documentation system detailing modeling andsimulation of NPP ESSI problems.
NRC ESSI Program
I Library Centric design and management approach (libraries: CompGeoMech UCD libs,MOSS, PETSc, UMFPACK, LAPACK, BLAS, Ltensor)
I Application Programming Interface (API) documented in detailI Finite elements: Solids (single phase (u), fully coupled, porous solid – pore fluid, (u-p-U,
u-p)), structural (truss, beams, shells), special elements (gap, contact, gap–contact,seismic isolators: frictional, rubber, lead-rubber)
I Material models: linear and nonlinear elastic, isotropic and anisotropic, elastic-plastic(von Mises, Drucker-Prager, Mohr-Coulomb, SANIsand, Dafalias-Manzari, Pisano-EP(calibrated from G/Gmax curves), all can be perfectly plastic, isotropic hardening orsoftening and/or translational and/or rotational kinematic hardening (A-F type);
I Staged loading (self weight, service loads, hazard loads, seismic loads, aftershocks, etc.)I Seismic Input, analytic, using Domain Reduction Method, incorporates seismic body (S,
P) and surface (Rayleigh, Love) wavesI Buoyancy effects (fluid pressure on embedded foundation) modeled analytically (with full
coupling with pore fluid pressures changes in foundation soil/rock)I Parallel computations using Plastic Domain Decomposition method, for efficient run
time computational load balancing (for inelastic computations)I Domain Specific Language for model input
Verification and Validation
The NRC ESSI Program has an extensive (and ever growing) verification and validationsuite. This ensures the highest possible quality of numerical prediction results.
I Verification: a Mathematics issue, provides evidence that the model is solved correctly.I Validation: a Physics issue, provides evidence that the correct model is solved.I Prediction: use of computational model to foretell the state of a physical system under
consideration under conditions for which the computational model has not beenvalidated.
NRC ESSI Computer
I NRC ESSI Computer is a distributed memory parallel(DMP) computer designed for high performance, parallelfinite element simulations using NRC ESSI Program
I Multiple performance CPUs and Networks (on-board,InfiniBand, GigaBit) can be effectively used
I Most cost-performance effective design (BeoWulf)I NRC ESSI Program runs most efficiently on the
NRC ESSI Computer, however, it is also sourcecompatible with any DMP supercomputers (any large national supercomputer or anyother cluster parallel computer)
I Industry standard Message Passing Interface (MPI) for communication between computenodes
I Current setup: 208 CPU cores, 288 GB RAM, 24TB disk space, triple network(on-board, Gigabit and InfiniBand)
I By Spring 2013: 592 CPU cores, 800 GB RAM, 40TB disk space, triple networkI By Fall 2013: 784 CPU cores, 1056 GB RAM, 48TB disk space, triple network
NRC ESSI Notes
A hypertext documentation system describing in detail modeling and simulations of NPPESSI problems:
I Theoretical and Computational Formulations: Finite Element Methods, Elasto-Plasticity,Static and Dynamic Solutions, Parallel Computing
I Software and Hardware Platform: Object Oriented Design, Library Centric Design,Application Programing Interface (API for the source code), Domain Specific Language(DSL, C and English language bindings), Software Build Process, Hardware Platform
I Verification and Validation (V&V): Verification: code, components (elements,algorithms), static and dynamic solution advancement, wave propagation, EarthquakeSoil Structure Interaction; Validation: material modeling, wave propagation, EarthquakeSoil Structure Interaction
I Application to Practical Nuclear Power Plant Earthquake Soil/Rock Structure InteractionProblems (case studies):. ESSI with 3D, inclined, uncorrelated seismic waves,. ESSI with nonlinear soil/rock and foundation slip,. ESSI with seismic base isolators, horizontal and vertical isolation benefits and
detriments,. ESSI with nonlinear soil/rock, base isolators and nonlinear structural response,. ESSI analysis for high frequency response,. ESSI for design basis earthquakes, full nonlinear response (from the list above), safety
margins determination,. ESSI for large, beyond design basis earthquakes, full nonlinear response (from the list
above), safety margins determinationI Draft of the NRC ESSI Notes is available at: http://nrc-essi-simulator.info/
Stochastic Modeling
In addition to deterministic modeling and simulations, recently developed ProbabilisticElasto-Plasticity (PEP) and Stochastic Elastic-Plastic Finite Element Methods(SEPFEM) (Jeremic et al.), are currently being integrated into the NRC ESSI Simulator.This will allow for very accurate (analytic, using solution to Fokker – Planck –Kolmogorov equation for PEP) development of full Probability Density Functions (PDFs)for generalized displacements, stress, strain, etc. From developed PDFs one can easilydevelop full Cumulative Density Functions (CDFs) aka fragility curves for anypart/components of the NPP soil/rock-structure system.
Verification Example (select)
Verification: Inclined (at 45o), 3D, body and surface seismic wave field:Models Stress Drop Disp. H↓ V↑ Acc. H↓ V↑ Radiation H↓ V↑
0m 5000m 10000m
5000m
0m
2000m
2000
m
Fault
-0.005
-0.004
-0.003
-0.002
-0.001
0
0.001
0.002
0.003
0 1 2 3 4 5 6
Dis
plac
emen
t (m
)
Time (s)
-0.0004
-0.0003
-0.0002
-0.0001
0
0.0001
0.0002
0.0003
0.0004
0 2 4 6 8 10 12
Dis
plac
emen
t (m
)
Time (s)
DRMFault Slip Model
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0 2 4 6 8 10 12
Acc
eler
atio
n (m
/s2 )
Time (s)
DRMFault Slip Model
-0.0004
-0.0003
-0.0002
-0.0001
0
0.0001
0.0002
0.0003
0.0004
0 2 4 6 8 10 12
Dis
plac
emen
t (m
)
Time (s)
0m
0m
70m
240m
DRM Layer
0
2e-05
4e-05
6e-05
8e-05
0.0001
0.00012
0.00014
0.00016
0 2 4 6 8 10
Ric
ker2
nd F
FT
Am
plitu
de
Frequency (Hz)
-0.0004
-0.0003
-0.0002
-0.0001
0
0.0001
0.0002
0.0003
0.0004
0 2 4 6 8 10 12
Dis
plac
emen
t (m
)
Time (s)
DRMFault Slip Model
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0 2 4 6 8 10 12
Acc
eler
atio
n (m
/s2 )
Time (s)
DRMFault Slip Model
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0 2 4 6 8 10 12
Acc
eler
atio
n (m
/s2 )
Time (s)
Nonlinear Example (select)
Gapping–Slipping at the foundation–soil/rock contact, for full 3D (at 45o, horizontal andvertical, body and surface) seismic wave:
Model and Input Acc. H ↑ and FFT ↓ Acc. V ↑ and FFT ↓
-40
-20
0
20
40
0 1 2 3 4 5 6 7 8
Acc
eler
atio
n (m
/s2)
Time (s)
Slip BehaviorNo-slip Behavior
-40
-20
0
20
40
0 1 2 3 4 5 6 7 8
Acc
eler
atio
n (m
/s2)
Time (s)
Slip BehaviorNo-slip Behavior
-4
-3
-2
-1
0
1
2
3
4
5
6
0 0.5 1 1.5 2 2.5 3 3.5 4
Acc
eler
atio
n (m
/s2 )
Time (s)
0
1
2
3
4
5
6
0 5 10 15 20
FF
T A
mpl
itude
(m
s)
Frequency (Hz)
Slip BehaviorNo-slip Behavior
0
1
2
3
4
5
6
0 5 10 15 20
FF
T A
mpl
itude
(m
s)
Frequency (Hz)
Slip BehaviorNo-slip Behavior
Gaping Response
4.5s 4.6s 4.7s
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
4.8s 4.9s 5.0s
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
5.1s 5.2s 5.3s
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.02
0.04
0.06
0.08
0.1
0.12
Slipping Response
4.5s 4.6s 4.7s
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
4.8s 4.9s 5.0s
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
5.1s 5.2s 5.3s
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
30 40 50 60 70 80 90 100 110
X (m)
30
40
50
60
70
80
90
100
110
Y (
m)
0
0.005
0.01
0.015
0.02
0.025
0.03
Current Example Problem
Representative NPP, emphasis on very accurate modeling and simulation of:
I Body and surface seismic wave effects,I Elastic-plastic soil/rock and structures,I Inelastic contact/gap (foundation-soil)I Seismic isolator effectsI Buoyant effects for embedded foundationsI Seismic wave frequencies ≤ 50HzI High Fidelity model: soil/rock block:
230m × 230m × 100m, foundation slab 90m × 90m,Containment Structure: 40m × 50m, 2.1 Million DOFs, 700,000 elements,
NRC ESSI Simulator License
A version of Attribution-NonCommercial-ShareAlike 3.0 Unported (CCBY-NC-SA 3.0) license, for components and the system.
Education is the Key!
Available are: short courses, lectures, advising and consultancy, etc.
Current and Former Student Collaborators and Developers
I Drs: Guanzhou Jie, Matthias Preisig, Zhao Cheng, Nima Tafazzoli, Federico Pisano;I Professors: Zhaohui Yang, Kallol Sett, Mahdi Taiebat;I Students: Panagiota Tasiopoulou, Babak Kamranimoghadam, Chang-Gyun Jeong, Jose
Abell Mena, Benjamin Aldridge, Justin Anderson;
Acknowledgement
The NRC ESSI Simulator System is developed in close collaboration and with fundingfrom the U.S. NRC. Such support is greatly appreciated. Additional partial funding fromCNSC/CCSN, US DOE and US NSF is also much appreciated.
Created with LATEXbeamerposter http://nrc-essi-simulator.info/
For more Information contact: Boris Jeremic, T 1.530.754.9248; k [email protected] or k [email protected] http://nrc-essi-simulator.info/