Preview of the New AIR South America Earthquake Models
CONFIDENTIAL ©2015 AIR WORLDWIDE 1
CONFIDENTIAL ©2015 AIR WORLDWIDE 2 Agenda
• New Hazard Modeling: Data & Methods • New Vulnerability Modeling: Data & Methods • Model Validation: Component-Level & Loss Estimation • Software: New Features for South America
CONFIDENTIAL ©2015 AIR WORLDWIDE 3 New Hazard Modeling
Mesut Turel
CONFIDENTIAL ©2015 AIR WORLDWIDE 4 South America Is One of the Most Seismically Active Regions of the World
Nazca Plate South America Plate
http://sicarius.wr.usgs.gov/IDB/
2007 Pisco
2014 Iquique
2010 Maule Earthquake
2010 Maule 1960 M9.6 Earthquake Length: 1000 km 1960 M9.6*
Width: 100 – 150 km Maximum Uplift: ~ 5.7 m
Maximum Subsidence: ~ 2.3 m
* Largest recorded earthquake worldwide
CONFIDENTIAL ©2015 AIR WORLDWIDE 5 The AIR Earthquake Models for South America Is Receiving a Comprehensive Update
Industry Exposure Database New Shake Catalog Damage Seismicity Time-Dependent Function Model Methodology AIR Intensity New Risk Earthquake Types Modeling Models for GMPEs Industry Facilities, Soil Maps Infrastructure, Builder’s Risk South America
Liquefaction New Seismic Risk Design Codes Tsunami Risk on the Pacific Coast
CONFIDENTIAL ©2015 AIR WORLDWIDE 6 Stochastic Event Generation Requires Latest Data Sources and Knowledge
- Historical earthquakes, fault • Historical earthquake parameters, and GPS data constrain catalog data the magnitude-rate in a seismic • Fault slip rates source zone • GPS strain rates
Seismic moment rate Driven by Driven by: • Historical earthquake • Historical catalog (Seismic budget) catalog for M≥4 from • Fault slip rates 1471 - 2014 • GPS data
Background seismicity on Characteristic earthquakes unknown faults and on known faults
major fault systems Cumulative Rate Cumulative
Magnitude Seismic Source Zone Model
CONFIDENTIAL ©2015 AIR WORLDWIDE 7 The History of Past Events Is of Great Importance for Characterizing the Seismic Hazard
• Extensive evaluation of data sources including work of the Global Earthquake Model
• Homogenization of data to moment magnitude (Mw) scale • Compilation of unified, comprehensive catalog of 47k events Mw≥4 from 1471 to 2014 based on a catalog quality and magnitude scale
CONFIDENTIAL ©2015 AIR WORLDWIDE 8
Characterizing Subduction Zone Locking and Deformation
Subduction Zone Locking Pre-2010 Maule GPS Stations
GPS Stations
Creeping Subduction Zone
Locked Subduction Zone
CONFIDENTIAL ©2015 AIR WORLDWIDE 9 There Is a Need for Multi-Mode Time-Dependent Rupture Probability Models for Subduction Zones
Historical Earthquake Historical Epicenters Rupture Areas
CONFIDENTIAL ©2015 AIR WORLDWIDE 10 Standard Versus New Approach to Time-Dependent Modeling
Stochastic Renewal Model View New Time-Dependent Model View
3
2
1,2,3 1
Earthquakes Earthquakes
Stress Stress On Area On Cell 1 2 3 1 2 3 Time Time
CONFIDENTIAL ©2015 AIR WORLDWIDE 11 Advantages of the AIR Time-Dependent Rupture Probability Model
Earthquake Scenarios Likelihoods of earthquakes are quantified using physical data:
• State of locking on the subduction interface
• Impact of historical earthquakes
AIR time-dependent model can capture State of locking from complex rupture dynamics kinematic modeling
CONFIDENTIAL ©2015 AIR WORLDWIDE 12 Ground Motion Prediction Equations (GMPEs) Are Updated Using Latest Research
Site Effects
Path Effects Source [ ] Weighting factor
2007 Tocopilla Modeled PGA
CONFIDENTIAL ©2015 AIR WORLDWIDE 13 Most Recent Geological Maps and Microzonation Studies Are Used To Create Soil Maps
Santiago Bogota Lima Guayaquil Caracas
CONFIDENTIAL ©2015 AIR WORLDWIDE 14 The Need for a Tsunami Model Was Greatly Felt After the Destructive 2010 Maule Earthquake
BEFORE AFTER
CONFIDENTIAL ©2015 AIR WORLDWIDE 15 Numerical Modeling Best Captures Tsunami Complexities
schematic domain configuration for an event
Constitución
Dy
Qi,j
Pi,j
Zi,j
Domain Resolutions 132 m 925 m Dx 6475 m
CONFIDENTIAL ©2015 AIR WORLDWIDE 16 Liquefaction Risk Is Modeled Explicitly
Liquefaction Hazard Maps Surficial Geology Maps Groundwater Depth Maps
CONFIDENTIAL ©2015 AIR WORLDWIDE 17 New Vulnerability Modeling
CONFIDENTIAL ©2015 AIR WORLDWIDE 18 Vulnerability Update Uses State-of-the-Art Engineering and Data for Damage Estimation
Risk Type Shake Tsunami Liquefaction
Building/Content/BI New New
Industrial Facilities New New New Infrastructure New New New
Builder’s Risk New New New
Auto New New Existing model
• New methodology for shake damage estimation
• Building code–based vulnerability classification for each country
CONFIDENTIAL ©2015 AIR WORLDWIDE 19 The Updated Damage Functions Are Generated Through Extensive Engineering Analyses
Subjected to 6900 Ground Motion Records from Earthquakes Worldwide
Representative Building Numerical Models Nonlinear Dynamic Analysis
0.35
0.3 Bounds of 0.25 viable damage 0.2 functions
MIDR 0.15
0.1
0.05 Mean Damage Ratio Damage Mean 0 0 0.5 1 1.5 2 Sa (g) 1.0 Intensity Estimate Losses Based on Building Building Response vs. Ground Motion Response Using PACT – FEMA P-58 Damage Ratio vs. Ground Motion
CONFIDENTIAL ©2015 AIR WORLDWIDE 20 Stringency of Seismic Design Code Is an Implicit Measure of Seismic Resistance in Vulnerability Assessment
Seismic Code Levels to Classify Vulnerability in AIR Model Vul. Class Description (code level) Without seismic consideration, mostly Pre refers to non-engineered buildings I Low With minimal seismic consideration II I Moderate II With moderate seismic consideration III I High II With stringent seismic consideration III I
II With very stringent seismic Special II consideration IV
CONFIDENTIAL ©2015 AIR WORLDWIDE 21 AIR Analysis of Building Code Evolution in Colombia Was Reviewed by Local Experts
1967 1983 1999 M6.8 M5.6 M6.2 Huila Earthquake Popayan Earthquake Armenia Earthquake
CCCSR-84 - 1984 NSR 98 - 1998 NSR 10 - 2010 noncompliance legal penalties Advances the incorporation of the More detailed construction class SEAOC Translation - 1974 included soil’s effects. and soil specification First seismic zonation map AIS 100 81 - 1981 First Colombian Code
ATC-3-06 Translation - 1979
Pre 1974 1975-1979 1980 -1984 1985 -1998 1999 -2010 Post 2011
CONFIDENTIAL ©2015 AIR WORLDWIDE 22 Regional Variation in Vulnerability Is Considered Using Seismic Design Zonation
CONFIDENTIAL ©2015 AIR WORLDWIDE 23 AIR Model Captures the Temporal and Spatial Variation of Vulnerability by Incorporating the Code Evolution
CONFIDENTIAL ©2015 AIR WORLDWIDE 24 Complex Industrial Facilities Are Modeled Using Dedicated, Component-Based Damage Functions
• Power Plants • Heavy Fabrication and • Chemical Processing • Water Systems Assembly • Metal and Mineral • Gas Processing • Light Fabrication and Processing Systems Assembly • High Technology • General Building/ • Food and Drug • Mining Construction Contractors Processing • Oil Refineries
CONFIDENTIAL ©2015 AIR WORLDWIDE 25 The Updated Models Support Infrastructure Systems with an Improved Set of Vulnerability Functions
CONFIDENTIAL ©2015 AIR WORLDWIDE 26 The Updated Model Supports the Builder’s Risk Line of Business
Risk for buildings under construction is characterized by time-variability of vulnerability and replacement
Phase I: Foundation, Phase II: Superstructure Phase III: Exterior Phase IV: Building Finishes, Substructure (Framing and Decking) Wall, Door & Windows Mechanical, Electrical Construction time is broken down to four distinct phases
CONFIDENTIAL ©2015 AIR WORLDWIDE 27 Ground Settlement Is the Salient Parameter in Estimating Liquefaction Damage
Empirical relationships provide an estimate of ground displacement using the ground motion parameters and soil properties
Settlement
AIR damage survey of Christchurch 2011
AIR damage functions are developed AIR by leveraging existing studies and HAZUS using observational data from Japan and New Zealand earthquakes Ratio Damage
Ground Settlement CONFIDENTIAL ©2015 AIR WORLDWIDE 28 AIR Tsunami Vulnerability Model Accounts for Three Damage Determinants
Velocity
San Antonio, 2010 Maule Earthquake Hydrostatic force Drag force
Equivalent inundation
Port debris observed after 2010 Maule Earthquake
CONFIDENTIAL ©2015 AIR WORLDWIDE 29 Validation and Software
Mark Szretter
CONFIDENTIAL ©2015 AIR WORLDWIDE 30 Time-Dependency of AIR’s Catalog Should Validate With the Energy-Release Observed in Maule 2010 Event
Magnitude-Frequency Plot: Region of the 2010 Maule Earthquake
TD Stochastic ≥M8.0 2010 Maule Rupture
CONFIDENTIAL ©2015 AIR WORLDWIDE 31 Time-Dependency of AIR’s Catalog Should Validate With the Absence of ‘Great’ Event in Peru Since 1746
Magnitude-Frequency Plot: Region Near Lima, Peru
Locked (red) subduction zone In the region around Lima, Peru (designated area in yellow)
CONFIDENTIAL ©2015 AIR WORLDWIDE 32 Modeled Ground Motion Intensities Match Observations in the Major Affected Provinces in the 2007 Pisco Event
Chincha Province Pisco Province Ica Province
Reported Modeled Reported Modeled Reported Modeled Location Location Location MMI MMI MMI MMI MMI MMI
Chincha Alta 7.5 7 San Clemente 7.5 8 Ica 7 7
Tambo de Mora 7 7.5 Pisco 8 8 Santiago 7 7
Chincha Baja 7.5 7.5 San Andres 7.5 7.5 Tate 6.5 7
Paracas 7.5 7.5 CONFIDENTIAL ©2015 AIR WORLDWIDE 33 AIR Carefully Validated Tsunami Extents and Maximum Heights Against Historic Event Data
2010 Maule Tsunami - City of Constitución Maule 2010 Observed Depths vs. Modeled
Observed Extent AIR Model Extent
30 Max ReportedObserved Modeled 20 Max Modeled
10
0
Max. Inundation Validation for Tsunami Historical Events
CONFIDENTIAL ©2015 AIR WORLDWIDE 34 AIR Participated in a South American Seismology Summit and Collaborated with Local Experts
• Dr. Diana Comte, Departamento de Geología y Geofísica, Facultad de Ciencias Físicas y Matemáticas, Universidad de Chile
• Dr. Carlos A. Vargas, Profesor Asociado en Departamento de Geociencias, Universidad Nacional de Colombia
• Dr. Daniel Carrizo, Departamento de Geología y Geofísica, Facultad de Ciencias Físicas y Matemáticas, Universidad de Chile
CONFIDENTIAL ©2015 AIR WORLDWIDE 35 Vulnerability of South American Construction Revised and Validated Against Recent Event Observations and Local Studies
New LocalDamage Studies Functions Suggest Align CM Better and RM with Less Recent Vulnerable Event-based Relative Observations to URM
0.50
0.45 Confined Masonry - AIR
0.40 Observation, Lower 0.35
0.30 Observation, Upper
0.25
0.20 Mean Damage Ratio Damage Mean
0.15
0.10
0.05
0.00 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 SA 0.3 sec
CONFIDENTIAL ©2015 AIR WORLDWIDE 36 AIR Considers Multiple Data Sources for Validation Including Local Expertise and Global Best Practices
Prof. Luis Yamín Universidad de Los Andes, COLOMBIA
Prof. Fabricio Yépez Universidad San Francisco de Quito, ECUADOR
Prof. Juan Carlos de la Llera Martin Pontificia Universidad Católica de CHILE
Prof. José Grases Universidad Central de VENEZUELA
Prof. Jorge Olarte Universidad Nacional de Ingenieria, PERU
CONFIDENTIAL ©2015 AIR WORLDWIDE 37
New 1Km Resolution Industry Exposure Databases Allows AIR to Better Validate Model Loss Estimates
Exposure Data Collection Risk Counts
• Population & Valparaiso Economic Censuses • Housing Surveys
Risk Attributes Santiago • Occupancy Melipilla • Construction & Height • Floor Area
Total Exposure
Total Replacement Value
Construction Costs Rancagua
• Costing Manuals • Construction Indexes
CONFIDENTIAL ©2015 AIR WORLDWIDE 38 Historic Event Losses for Significant Earthquake Events Provide Important Guide Points for Validation
Economic Losses for Colombian Earthquake Events 1990-Present
Ground Up Loss Estimates Loss Up Ground Estimates Loss Up Ground
1995 1994 1992 1999 Cali Páez Antioquia Zona Cafetera Armenia - Quindío
CONFIDENTIAL ©2015 AIR WORLDWIDE 39 Model’s Loss Distribution by Region Matches the Regional Losses from Reports – 2007 Pisco
Modeled and Reported Contributions to Loss at Department Level 90% Reported Contribution Model Contribution 80% Source: INEI (2007), Census of the Areas Affected by the Earthquake 70%
60%
50%
40%
30%
20%
10%
0% ICA LIMA HUANCAVELICA
CONFIDENTIAL ©2015 AIR WORLDWIDE 40 Model Correctly Reproduces the Pisco Earthquake Observed Loss Ratios at the Province Level – Chincha, Peru
Modeled Ground Chincha Province Motion *Spectral Acceleration Sa 0.3 sec (g) Modeled vs. Observed Loss Ratios Chincha Province
Observed Observed AIR District Damage Damage Modeled Lower Bound Upper Bound Damage
18% 35% 35% ALTO LARAN 25% 43% 35% CHINCHA ALTA 27% 46% 30% CHINCHA BAJA 20% 39% 33% EL CARMEN 28% 48% 41% GROCIO PRADO 29% 48% 40% PUEBLO NUEVO 30% 50% 38% SUNAMPE 26% 45% 32% TAMBO DE MORA
CONFIDENTIAL ©2015 AIR WORLDWIDE 41 2010 M8.8 Event in Concepción, Chile Provided Good Data Source of Claims Data for Validation
CONFIDENTIAL ©2015 AIR WORLDWIDE 42 Touchstone’s Flexible Options Allow for Separate Evaluation of Tsunami and Shake Losses and Improved Disaggregation
CONFIDENTIAL ©2015 AIR WORLDWIDE 43 AIR’s Touchstone Platform Allows Analysis of Exposure Distribution and Deterministic Studies Relative to Hazard
CONFIDENTIAL ©2015 AIR WORLDWIDE 44 AIR’s New South America Earthquake Models Set the Bar in Terms of Scope, Innovation, and Quality
• Tsunami modeling for the Pacific Coast subduction zone from Chile to Colombia and liquefaction • New risk types: industrial, public infrastructure, CAR builder’s risk • Extensive validation of components, loss outputs AIR Model Coverage using real loss data and with inputs from local engineers and geo-scientists in each of the modeled countries
CONFIDENTIAL ©2015 AIR WORLDWIDE 45 Additional Resources on the AIR Website
April 8th – 10th in Boston
CONFIDENTIAL ©2015 AIR WORLDWIDE 46