Department of Mechanical Engineering, Tsinghua University Beijing, RPC, July 7, 2006

INTEGRATED EARTHQUAKE SIMULATION INTEGRATED EARTHQUAKE SIMULATION

M. HORI Earthquake Research Institute, University of Tokyo

CONTENTS Full computation of earthquake hazards and disasters Examples of integrated earthquake simulation

Recent Achievement on Earthquake Research Recent Achievement on Earthquake Research

Computation of Earthquake Wave Propagation
Prof. Furumra (ERI, U. Tokyo): use of Earth Simulator

Dynamic structure analysis methods used for design non-linear due to materials and large-deformation

Development of GIS Storing Structure Data

OVERVIEW OF IES OVERVIEW OF IES

computer model of city
houses/buildings public spaces lifelines infrastructures

structure response simulation

earthquake
amplification

action against earthquake

crisis management time propagation evacuation retrofitting recovery

fault mechanism

earthquake simulation

action simulation

SYSTEM OF IES SYSTEM OF IES

IES earthquake
earthquake scenario SGM distribution fault mechanism propagation amplification

a set of (application) layers of GIS

structure response
model construction disaster prediction steel structure concrete structure geotechnology buildings/architecture data: underground results: SGM data: structure results: damage

GIS

action
people agents community society

MACRO-MICRO ANALYSIS METHOD MACRO-MICRO ANALYSIS METHOD

stochastic modeling of underground structure

bounding medium theory

multi-scale analysis pessimistic optimistic

macro-analysis
geological scale

micro-analysis
engineering scale

EXAMPLE OF MACRO-MICRO ANALYSIS: EXAMPLE OF MACRO-MICRO ANALYSIS: YOKOHAMA CITY SIMULATION YOKOHAMA CITY SIMULATION
Comparison of synthesized strong ground motion with data observed at 13 seismograph sites
139.5E 140.0E 140.5E

as06

Yokohama City
35.5N

35.5N

hd06

epicenter

is06
0 35.0N 139.5E 140.0E

km 7.5 15 35.0N 140.5E

August 11, 1999 Lat. Long. Depth Strike 35.4N 139.8E 53km 62

Dip

Rake Mag. 85 73 4.0Mw

MODEL FOR MICRO-ANALYSIS MODEL FOR MICRO-ANALYSIS

N
139.5E 140.0E 140.5E

E 40[km]

35.5N

35.5N

70[km]
35.0N 139.5E 140.0E 35.0N 140.5E

30[km]
p w ave veloc.[m /sec] s w ave veloc.[m /sec] density[kg/m ]
3

m ate.1 1040 600 1800

m ate.2 1730 1000 2000

m ate.3 2950 1700 2300

m ate.4 5200 3000 2500

mate.1 mate.2

mate.3 mate.4

between 1st and 2nd layers

0 [m]

between 2nd and 3rd layers

[m] 0 -1250 -2500 0

between 3rd and 4th layers

[m]

-500 -1000

-2500 -5000

MODEL FOR MICOR-ANALYSIS MODEL FOR MICOR-ANALYSIS

400[m] N 400[m] as06
[m] 10 20 30 40 50 401 402 403 404 405 414 layer Ac1 Ac2 As2 T properties of layer [kg/m] 1500 1600 1800 2100
3

[m/sec] 100 140 210 700

model of site as06s

400[m] 400[m]

b[m/s] 600 500 400 300 200 100

50[m]

model 1 (optimistic)

model 2 (pessimistic)

MACRO-ANALYSIS RESUTLS MACRO-ANALYSIS RESUTLS

velocity [kine] 0.08 upper bounding medium 0.04 0.00 0 2 4 6 lower bounding medium measured

a) case 1
velocity [kine] 0.02 0.01 0.00 0 2 4 6

8 time [sec]

simple fault mechanism used accuracy guaranteed up to 2[Hz]

b) case 2

8 time [sec]

agreement with observed data

MICRO-ANALYSIS RESULTS (1) MICRO-ANALYSIS RESULTS (1)

meas. north/south
100 amplitude[cm]

opt.

pes. east/west up/down

10

1 0.25

1.25

2.25

3.25

4.25 frequency[Hz]

PGV [kine] SI

meas. opt. 0.170 0.264 0.191 0.318

pes. 0.292 0.339

PGV [kine] SI

meas. 0.234 0.295

opt. 0.316 0.348

pes. 0.391 0.415

PGV [kine] SI

meas. 0.049 0.053

opt. 0.093 0.073

pes. 0.133 0.101

fair agreement with observed data

MICRO-ANALYSIS RESULTS (2) MICRO-ANALYSIS RESULTS (2)

max. velocity[kin] 1

lower bounding medium

1

0.1

0.1

0.01

measured

upper bounding medium

0.01 site number

a) case 1
2 sites have consistent errors, probably due to poor modeling

b) case 2 fair agreement with observed data

SIMULATION USING GIS DATA SIMULATION USING GIS DATA

Structure Data Stored in GIS
basic properties (location, owner) configuration structure type

Construction of Models
presumption of standard material and structure properties standard structure model

Automatic Construction
automatic construction of GIS data to standard structure model highest redundancy needed

EARTHQUAKE DISASTER SIMULATION EARTHQUAKE DISASTER SIMULATION DEVELOPMENT OF VIRTUAL TOWN/CITY DEVELOPMENT OF VIRTUAL TOWN/CITY
size [m] virtual town virtual city 1 virtual city 2 virtual city 3 300x300 1,000x2,000 400x440 300x300 data GIS GIS with CAD GIS with CAD GIS with AI earthquake MMAM MMAM MMAM MMAM structure SDOF for building VFEM for buildings VFEM for infrastructures action none none none

FEM based on Multi-Agent ERD Simulation

VFEM Voxel Finite Element Method MMAM Macro-micro Analysis Method ERD Earthquake Resistant Design

VIRTUAL TOWN VIRTUAL TOWN

GIS for bore holes (surface layers)

Roppongi Area: 300x300[m]

GIS for buildings

DATA AND COMPUTER MODEL DATA AND COMPUTER MODEL

boring site
F D E C A

B
N

target area (300x300[m])

60[m]

300[m] 300[m]
number of layer 1 2 3 4 5 bottom soil type surface soil loam sand clay fine sand rock density(g/cm ) 1.625 1.550 1.800 1.750 1.900 1.850
3

147 buildings
Sv (m/s) 120.0 135.0 400.0 200.0 425.0 600.0 Pv (m/s) 204.0 229.5 680.0 340.0 722.5 1020.0

Materials Wooden buildings R.C. S.R.C.

Fundamental Period varies from 0.2 sec. to 0.7 sec. T=0.02H T=0.03H

SIMULATION OF VIRUTAL TOWN SIMULATION OF VIRUTAL TOWN

high spatial resolution

large variability in response

VIRTUAL CITY 1 VIRTUAL CITY 1

1km 2km

data CAD data for all buildings borehole data for underground

CAD data borehole data

RESPONSE OF BUILDINGS IN VIRTUAL CITY 1 RESPONSE OF BUILDINGS IN VIRTUAL CITY 1

high frequency dominant

low frequency dominant

HIGH RESOLUTION OF IES HIGH RESOLUTION OF IES

part of virtual city

room

VIRTUAL CITY 2 VIRTUAL CITY 2

data CAD data for all buildings borehole data for underground

RESPONSE OF INFRASTRUCTURES AND RESPONSE OF INFRASTRUCTURES AND BUILDINGS IN VIRTUAL CITY 2 BUILDINGS IN VIRTUAL CITY 2

RESPONSE OF VIRTUAL CITY 2 RESPONSE OF VIRTUAL CITY 2

CLOSE-UP VIEW OF STRUCTURE: CLOSE-UP VIEW OF STRUCTURE: DYNAMIC RESPONSE OF BRIDGE 1 DYNAMIC RESPONSE OF BRIDGE 1

X direction

Y direction

Z direction

VIRTUAL CITY 3 VIRTUAL CITY 3

soil layer concrete pier steel pier

300[m]

ground mold

gas pipe line

300[m]

PLUG-IN OF NUMERICAL METHODS PLUG-IN OF NUMERICAL METHODS

PLUG-IN: request programmer to modify source codes such that I/O can be made through common protocols heavy task for programmer due to variety of structure data

Methodology of Integration
Application of DataBase management techniques Federation-type DataBase with aid of interpreting programs (agents) interpreting program: supports data exchange between kernel and each simulation program
interpreter program A

IES

SP-D simulation program SP-A SP-C SP-B

MEDIATOR AS INTERPETING AGENT MEDIATOR AS INTERPETING AGENT

IES structure with mediator
take location data GIS library visualization VRML visualize UGSM micro-analysis analysis kernel take data select point analyze response simulation program put make VRML take response analysis mediator

major role of mediator 1. data exchange 2. execution of program

INHERETANCE OF MEDIAOTR OBJECT INHERETANCE OF MEDIAOTR OBJECT

Program Structure of Mediator
structure node successi on element FEM material Gauss point successi on for structure A property location, property VRML A VRML POV AVS VMRL_A location, displacement data nodes, type, find structure take data from library take data select point select point to which strong motion is input

Methods
A find structure from GIS put take

object of mediator for structure A put data to simulation program take results of simulation program

visualize results of simulation program make VRML

run simulation program analyze response

Computational Science Approach: seek to automatic coding of mediator program by taking full advantage of inheritance

MEDIATOR MAKER AS AI MEDIATOR MAKER AS AI

Mediator Maker
artificial intelligence of automatically generating a mediator for a given simulation program mediator maker analyzes source code of a simulation program, and constructs a code of a corresponding mediator
analysis of source code subroutine common memory source codes

digested code

It is possible to make mediator maker by taking advantage of

common structure of source code (FEMbased) usage of common key variables (such as number of nodes, elements, etc.)

construction of mediator analysis of I/O analysis of loop and code for mediator condition prediction of variable

MODELS OF INFRASTRUCUTRES IN MODELS OF INFRASTRUCUTRES IN VIRTUAL CITY 3 VIRTUAL CITY 3

1.7x1.7 5x2.5 10 5 5 0.5 1 15

infrastructures gas pipe line concrete pier steel pier ground mold particular material properties & mechanism for earthquake resistance

3 1

3 10

20 8 14 1x1 10

8 1

reliable analysis methods used in ERD are plugged in IES

NON-LINEAR RESPONSE OF NON-LINEAR RESPONSE OF INFRASTRUCTURES IN VIRTUAL CITY 3 INFRASTRUCTURES IN VIRTUAL CITY 3

steel pier

ground mold

plugged-in simulation programs work in IES concrete pier

RESPONSE OF VIRTUAL CITY 3 RESPONSE OF VIRTUAL CITY 3

all numerical analysis of structures are non-linear based on earthquake resistant design

unified visualization will contribute to form common recognition of possible earthquake disaster

SNAPSHOT OF RESPONSE AND DAMAGE SNAPSHOT OF RESPONSE AND DAMAGE

SIMULATION OF ACTION SIMULATION OF ACTION

Evacuation Simulation
physical simulation multi-agent simulation model based on flow, potential, etc. intelligent agents

need to estimate panic in evacuation process

Damage and Loss Simulation

micro: evaluation of damage and repair cost macro: based on past record of damage risk analysis/management

prediction of damage and loss of future society

ACTION SIMULATION: ACTION SIMULATION: EVACUATION SIMULATION USING MAS EVACUATION SIMULATION USING MAS
Evacuation Simulation
predict evacuation process in underground towns, high-rise buildings, department stores, schools, etc. use intelligent agents to mimic people in panic state

underground subway station 4 floors 3 elevators

MULTI-AGENT SIMULATION MULTI-AGENT SIMULATION

Utilization of Intelligent Agent
input output internal variables (max. speed, intelligence, memory) external state see surrounding environment, agent, etc. think judge most suitable path move go to next position

Ability MaximumSpeed Visibility Intelligence Thought Direction Speed Path

Agent Ability Thought See(): Wall, Corridor, Agent Think(): Destination, Direction, Speed Move(): Direction, Check

MODELING FOR DYNAMIC STRUCTUREIS & MODELING FOR DYNAMIC STRUCTUREIS & EVACUATION ANALYSIS EVACUATION ANALYSIS

modeling strategy structure data for each floor data for escalator

evacuation path corridors determined from structure data Exit determined from escalator data structure is arranged to be used for structure analysis and for evacuation analysis

RESULTS OF SIMULATION (1) RESULTS OF SIMULATION (1)

dynamic structure analysis model structure & ground simulation MMAM

structure response analysis is made by accounting for soil-structure interaction effects

no damage

RESULTS OF SIMULATION (2) RESULTS OF SIMULATION (2)

upper floor escalator & stairs lower floor green less intelligent agent - slow think gray more intelligent agent - quick think

damage

multi-agent analysis is possible for structures to which dynamic structure analysis is made.

DIGITAL KOBE DIGITAL KOBE

Kobe City, Higashi-Nada District 500[m]x700[m]

70 0[

m]

buildings
500 [m]

bridge

1262 wooden, RC/SRC for highway 14 RC piers

SIMULATION RESULTS (1) SIMULATION RESULTS (1)

SIMULATION RESULTS (2): SIMULATION RESULTS (2): THREE EARTHAUEK SCENARIOS THREE EARTHAUEK SCENARIOS
Case 1 Case 1 Case 2

stress concentration Case 2 Case 3

Case 3 differences in earthquake disaster due to difference in earthquake scenario

FULL CITY SIMULATION FULL CITY SIMULATION

DIGITAL BUNKYO CITY DIGITAL BUNKYO CITY

ground structure model 700[m]700[m]40[m] 3 layers

building models around 2000 MDOF system

data source GIS of boring data of Tokyo Metro-police commercially available GIS for structure data

STRONG GROUND MOTION DISTRIBUION STRONG GROUND MOTION DISTRIBUION

Expected Kanto Earthquake

Recorded Hanshin-Awaji Earthquake

STRONG GROUND MOTION DISTRIBUTION STRONG GROUND MOTION DISTRIBUTION

Expected Kanto Earthquake

Recorded Hanshin-Awaji Earthquake

VIBRATION OF BUILDINGS VIBRATION OF BUILDINGS

VIBRATION OF BUILDINGS VIBRATION OF BUILDINGS

EVACUATION OF RESIDENCES EVACUATION OF RESIDENCES

-31600 -31800 -32000 -32200 -32400 -32600 -32800 -33000 -8000

-7500

-7000

road network model

simulation for various situations is possible.

multi-agent simulation

SIMULATION OF TOKYO STATION SIMULATION OF TOKYO STATION

EXAMPLE 1 EXAMPLE 1

strong ground motion

structure-wise damage evaluation

EXAMPLE 2 EXAMPLE 2

TOKYO DOWNTOWN TOKYO DOWNTOWN

EXAMPLE 1 EXAMPLE 1

scenario 1 strong ground motion

EXAMPLE 2 EXAMPLE 2

1 structure response

CONCLUDING REMARKS CONCLUDING REMARKS

Potentials of IES
Generation of new information on natural disaster risk and hazard by applying numerical simulation to data stored in GIS Improve residents awareness on natural disaster risk and hazards and form common recognition of them among community

Future Tasks
Advancement of structure data stored in GIS new exclusive sensing of buildings and structures Advancement of natural disaster preparedness, mitigation and recovery significant need for integration of technologies for natural disasters