Appln High-Performance Computing Earthquake-Related Problems

Contents
Preface
About the Editor
1. Earthquake Generation Cycle and Crust Deformation in Subduction Zones
	1. Introduction
	2. Earthquake Generation Cycle Simulations for Nankai Subduction Zone in Southwest Japan
		2.1. Formulation of earthquake generation cycle
			2.1.1. Stress variation with spatio-temporal variation of slip
			2.1.2. Constitutive relationship of fault and evolution law of strength
			2.1.3. Evaluation of the effect of stress changes due to surrounding earthquakes
		2.2. Numerical simulation method of earthquake generation cycle
		2.3. Problem setting
		2.4. Results
		2.5. Discussion
		2.6. Summary
	3. Calculation of Green’s Functions for the Subduction Zones in Japan
		3.1. Coordinate transformation between the geographic and a Cartesian coordinate system
		3.3. Calculation of Green’s functions for the subduction zones in Japan
		3.4. Application to slip deficit rate estimation in the Nankai Trough
		3.5. Summary
		3.6. Future directions
	Acknowledgment
	References
2. Large-Scale Simulation of the Discrete Element Method to Analyze Multi-Scale Granular Dynamics in the Practical Geoscience Applications
	1. Introduction
	2. Basics of DEM
		3.1. Efficient computation of preconditioning
		3.2. Kernel calculation with action–reaction law with many-core computing
		3.3. Performance test of shared memory parallelization methods
	4. Distributed Memory Parallelization
		4.1. MPI subdomains
		4.2. Iterative load-balancing method
		4.3. Multigrid level relaxation
		4.4. Flexible rectangular DD
		4.5. Verification of DEM implementation for the distributed memory system
		4.6. Validation test of iterative load-balancing scheme
		4.7. Parallel scalability test using the HPC machines
	5. Stress Chain Analysis
	6. Large-Scale DEM Application in Geoscience: Numerical Simulation of a Real-Scale Numerical Sandbox Experiment
		6.1. Horizontal undulation of thrust and arcuate stress state inside the granular layer
		6.2. Multi-scale mechanism for generating arcuate geological-scale structure
	7. Conclusion
	References
3. Multi-Scale Tsunami Simulation with High-Fidelity Modeling and Visualization
	1. Introduction
	2. Overview of a Stabilized ISPH for 3D Tsunami Inundation Analysis
		2.1. Incompressible smoothed particle hydrodynamics
		2.2. Spatial discretization by SPH
		2.3. Boundary treatment using a modified mirror boundary method
			2.3.1. Virtual markers with fixed wall particles in a structured regular grid
		2.4. Non-homogeneous pressure Neumann condition
		2.5. Validation with 3D dam break flow with the opening gate
	3. Matrix-Arrayed Virtual Wave Maker for 2D–3D Coupling Simulations
		3.1. General functions of the virtual wave maker
		3.2. VWM considering pressure from pre-analysis
		3.3. VWM with the SGFGP boundary treatment
		3.4. Validation test for VWM considering pressure from pre-analysis
	4. Pre- and Post-Processes for the Particle-Based Tsunami Inundation Simulation
		4.1. Preprocess for making a 3D high-fidelity model from Geographic Information System (GIS) data
		4.2. Post process of particle-based tsunami simulation data as a 3D photo-realistic visualization
			4.2.1. Smoothed surface representation from particles using the marching cubes method
			4.2.2. 3D filtering to decrease polygons
			4.2.3. Lattice size determination
	5. Numerical Convergence Tests for Tsunami Inundation Using the SPH
		5.1. Target city of tsunami inundation
		5.2. Simulation conditions
		5.3. Evaluation of resolution
		5.4. Discussion on the simulation results
		5.5. Photo-realistic visualization and virtual evaluation training system
	6. Validation — Representation of the 2011 Tohoku Earthquake and Tsunami at Utatsu
		6.1. 2D tsunami propagation analysis based on SWE (Level-0 analysis)
		6.2. Coupling between 2D shallow water (Level-0)–3D particle simulation (Level-1)
	7. Conclusion
	References
4. Seismic Structural Response Analysis for Extremely Strong Ground Motion Using Parallel Finite Element Method
	1. Introduction
	2. Trial Simulation
		2.1. Computation performance results
		2.2. Seismic structural response results
	3. Improvement of Parallel Finite Element Method
	4. Simulation of Extremely Strong Ground Motion
		4.1. Computation performance results
		4.2. Seismic structural response results
	5. Concluding Remarks
	Appendix A. Constitutive Relation Implemented into Parallel FEM
		A.1. Concrete constitutive relation
		A.2. Soil constitutive relation
		A.3. Model for contact surface between building and ground
	Appendix B. Capability of Liquefaction Analysis
	Appendix C. Capability of Lift-Up Analysis
	References
5. Earthquake Simulation on Unstructured Finite Elements Enhanced by High-Performance Computing
	1. Introduction
	2. Background
	3. CPU-Based Finite Element Solver for Nonlinear Wave Propagation
		3.1. Algorithm design
			3.1.1. Development of solver algorithm
			3.1.2. Development of finite element model construction method
		3.2. Performance
			3.2.1. Performance measurement environment
			3.2.2. Performance measurement setting
			3.2.3. Performance measurement results
		3.3. Example
	4. CPU-Based Finite Element Solver on Specialized Mesh for Nonlinear Wave Propagation
		4.1. Algorithm design
		4.2. Performance
			4.2.1. Performance measurement environment
			4.2.2. Performance measurement setting
			4.2.3. Weak scaling
			4.2.4. Strong scaling
			4.2.5. Performance on practical problem
		4.3. Example
	5. CPU-Based Finite Element Solver for Crust Deformation
		5.1. Algorithm design
		5.2. Performance
			5.2.1. Performance measurement environment
			5.2.2. Performance measurement setting
			5.2.3. Weak scaling
			5.2.4. Strong scaling
			5.2.5. Performance on Intel Xeon system
		5.3. Example
	6. CPU-Based Finite Element Solver for Nonlinear Soil–Structure System Response
		6.1. Algorithm design
			6.1.1. Details of solver algorithm
		6.2. Performance
			6.2.1. Performance measurement environment
			6.2.2. Problem measurement setting
			6.2.3. Time-to-solution
			6.2.4. Weak scaling
			6.2.5. Strong scaling
		6.3. Example
	7. GPU-Based Finite Element Solver for Nonlinear Soil–Structure System Response
		7.1. Algorithm design
			7.1.1. Enhancement of preconditioner by artificial intelligence
			7.1.2. Enhancement of preconditioner by transprecision computing
		7.2. Performance
			7.2.1. Performance measurement environment
			7.2.2. Performance measurement settings
			7.2.3. Performance measurement resul
		7.3. Example
	8. Conclusion and Future Tasks
	Acknowledgment
	References
6. Large-Scale Finite Element Simulation of Surface Faulting
	1. Introduction
	2. Numerical Analysis Method
		2.1. High-performance computing for fault displacement evaluation
		2.2. Three-dimensional joint element
		2.3. Hamiltonian and symplectic time integration
			2.3.1. Lagrangian and Hamiltonian
			2.3.2. Symplectic time integration
			2.3.3. Simple spring problem
		2.4. Implementation to parallel finite element program
			2.4.1. Nonlinear spring model on fault
			2.4.2. Hamiltonian for finite element method
	3. Verification of Numerical Analysis Program
		3.1. One-dimensional problem
		3.2. Three-dimensional problem
			3.2.1. Test one: Single-fault problem
			3.2.2. Test two: Multi-faults problem
	4. Application to Actual Earthquakes
		4.1. Basic concept of analytical model
		4.2. The 2014 Nagano-ken-hokubu earthquake
			4.2.1. Outline of the earthquake and surface faulting
			4.2.2. Analytical model
			4.2.3. Results
		4.3. The 2016 Kumamoto earthquake
			4.3.1. Outline of the earthquake and surface faulting
			4.3.2. Analytical model
			4.3.3. Results
		4.4. Discussion
			4.4.1. 2014 Nagano-ken-hokubu earthquake
			4.4.2. 2016 Kumamoto earthquake
	5. Predictive Simulation
		5.1. Setting of input conditions
			5.1.1. Basic slip distribution on primary faults
			5.1.2. Large surface slip area
		5.2. Results
			5.2.1. Discussion
	6. Example of Uncertainty Quantifications
		6.1. Uncertainty in fault displacement simulations
		6.2. Uncertainty of material properties
			6.2.1. Sensitivity analysis for material properties
			6.2.2. Probabilistic response and its convergence with Latin hypercube sampling
			6.2.3. Effects of each of the material properties and correlation between material properties with orthogonal sampling
		6.3. Uncertainty in fault planes
		6.4. Slip distribution across the earthquake source fault
		6.5. Geometry of secondary fault: Dip angle
		6.6. Number of secondary fault planes in analytical model
	7. Conclusion
	References
7. Traffic Flow Simulator and Travel Demand Simulators for Assessing Congestion on Roads After a Major Earthquake
	1. Introduction
	2. Motivations for the Study
		2.1. Past major traffic disruptions after major earthquakes in Japan
			2.1.1. 1995 Hanshin-Awaji earthquake case
			2.1.2. Disruption of Tokyo network in the 2011 Tohoku earthquake
			2.1.3. 2016 Kumamoto earthquakes case
			2.1.4. Lessons from three past earthquake cases
		2.2. Relevant existing studies of traffic and demand simulators
			2.2.1. Traffic simulators
			2.2.2. Demand simulators
	3. Outline of the Simulation System
	4. Traffic Flow Simulator — FastDUE
		4.1. Overview
		4.2. Models and calculation algorithms
			4.2.1. Traffic flow model
			4.2.2. Route choice model
		4.3. Asynchronized update and its parallelism
			4.3.1. Asynchronized update
			4.3.2. Forward loop and outer loop
			4.3.3. Parallelization
		4.4. Test cases
			4.4.1. Simple network
			4.4.2. Large-scale network
		4.5. Scalability
	5. Travel Demand Simulators — DeSuTA
		5.1. SPACE
		5.2. ASTRO
		5.3. HuLAND
	6. Case Study of Nankai Trough Earthquake
	Acknowledgments
	References
8. Dynamic Programming of Firms’ Activities and Market Interactions After a Disaster
	1. Introduction
	2. Model of Post-Disaster Economy
		2.1. Regions and markets
		2.2. Lifeline and supply-chain resumption probabilities, bifurcation of firm types
		2.3. Recovery investment problem for firms in Region OSK and Region ROK
		2.4. Production in Region ROJ
		2.5. Market and price change process
		2.6. Policy evaluation function
	3. Algorithm of Solving Bellman Equation and Simulating Dynamics of Economy
		3.1. Simulation steps
		3.2. Implementation and scalability
		4.1. Settings
		4.2. Results
	4. Case Study
	5. Conclusion and Future Tasks
	References
9. Toward the 1:1 Scale Agent-Based Simulation of Post-Disaster Economies
	1. Introduction
	2. Agent-Based Economic Models and HPC Implementation
		2.1. Agent-based economic models
		2.2. Agent types and their roles
			2.2.1. Firms
			2.2.2. Imports and exports
			2.2.3. Households
			2.2.4. Bank
			2.2.5. Central bank
			2.2.6. Central government
		2.3. Need for an HPC implementation
		2.4. Challenges in scalable HPC implementation
		2.5. Scalable high-performance implementation
			2.5.1. Scalable solutions for the centralized graphs
			2.5.2. Scalable solution for scale-free graphs
			2.5.3. Communication hiding
			2.5.4. Algorithmic improvements
		2.6. Computational performance
	3. Data Generation for Simulating the Japanese Economy
		3.1. Required parameters
		3.2. Data sources
		3.3. Industrial parameters
			3.3.1. Sector parameters
			3.3.2. Parameters of individual firms
			3.3.3. Foreign buyers’ and foreign sellers’ data
		3.4. Households’ data
		3.5. Capital consumption coefficients of firms and households
		3.6. Bank’s data
		3.7. Government’s data
		3.8. Central bank’s data
		3.9. Time histories data for firm agents’ production and price decisions
	4. Verification and Validation
		4.1. Verification
		4.2. Validation
			4.2.1. Problem settings
			4.2.2. Validation: The model with economy-wide forecasting for agents’ decision-making
			4.2.3. Validation: The model with sector-wise forecasting in firms’ decision-making
			4.2.4. Comparison of model performance under economy-wide and sector-wise forecasts
	5. Integration of the ABEM with Physics-Based Simulators
		5.1. Integrated earthquake simulator
		5.2. Estimation of repair costs
		5.3. Integration with the ABEM
	6. Demonstrative Simulation of a Post-Disaster Economy
		6.1. Modifications to the ABEM
			6.1.1. Firms
			6.1.2. Households
			6.1.3. Government
		6.2. Simulations of post-disaster economy
			6.2.1. Disaster and impacted region
			6.2.2. Infrastructure data from GIS
			6.2.3. Seismic performance evaluation using IES
			6.2.4. Loss estimation using PACT
			6.2.5. Post-disaster economic simulations using ABEM
	7. Concluding Remarks
	Acknowledgments
	References
10. Study on Improved and Advanced Urban Information Data for Implementation into Social Science Simulations of Earthquakes
	1. Introduction
	2. Development of Building Data
		2.1. Data
			2.1.1. GIS building data
			2.1.2. Street view image data
		2.2. Built year and structure prediction
			2.2.1. Construction year and structure estimation using building attributes
			2.2.2. Built year and structure prediction using street view
		2.3. Results
			2.3.1. Estimation of building structure material and year of construction using building attributes
			2.3.2. Building structure and construction year estimation using street view images
		2.4. Evaluating earthquake damage
	3. Development of People Flow Data
		3.1. Data
			3.1.1. GPS data
			3.1.2. Parson Trip data
		3.2. Development of people flow data
			3.2.1. Geocording to point-of-interest data
			3.2.2. Road network interpolation
			3.2.3. Scaling factor
			3.2.4. Estimation of attribute
			3.2.5. Results
		3.3. Validation
	4. Development of Business Transaction Economic Data
		4.1. Inter-company transaction data
		4.2. Mythology of development of business transaction economic data
			4.2.1. Estimation of number of employees of each establishment
			4.2.2. Estimation of transaction value
		4.3. Results
			4.3.1. Number of employees
			4.3.2. Transaction value
		4.4. Validation
			4.4.1. Validation check of number of employees
			4.4.2. Validation check of transaction value
	5. Advanced Simulations Through Data Linking
		5.1. Economic simulation
			5.1.1. Earthquake ground motion data
			5.1.2. Infrastructure damage
		5.2. Evacuation simulation
	6. Conclusion
	References
Index