2nd year SAFER Project meeting. Armada Hotel, Istanbul, Turkey 25-27 June, 2008. Information-dependent lead time maps for earthquake early warning in the.

2nd year SAFER Project meeting. Armada Hotel, Istanbul, Turkey 25-27 June, 2008.

Information-dependent lead time maps for earthquake early warning in the Campania region

Iunio IervolinoAssistant Professor of Structural Engineering

Dipartimento di Ingegneria Strutturale Università degli Studi di Napoli Federico II, Naples, Italy.

http://www.unina.it/

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi

onDone so far: RT- PSHA & Threshold set Uncertainty Analysis To doLead-Time Maps

Naples

Hospitals

Fire Stations

Lifelines

Factories

Railways

Highways

4 Cities

The Campanian case

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


Source-to-site distance

Seismic network

Ground motion at the site

IM (i.e. PGA)

Structural/non-structural performance/loss

EDP (i.e. Maximum Interstory Drift Ratio)

Epicenter

Signal at the network stations

Site-Specific Warning by Regional Networks

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


1 2 1 2| , ,..., | , ,...,| , | |

M R s s s

M R

f PGA f PGA m r f m f r s dr dm

PDF of magnitude conditioned to the measures of the seismic instruments

PDF of distance due to rapid localization method

Ordinary Attenuation relationship

Prediction of peak ground motion at the site

Real-Time Hazard Analysis for Hybrid EEWS

ν

ν

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


Real-Time loss assessment

Extending the hazard approach it is possible to determine the expected losses conditioned to the measurements of the seismic network in the

case of alarming or not

Expected Loss

1. Loss probability depending on the alarming decision

2. Structural damage probability depending on

building’s seismic response

3. Seismic response probability depending

on hazard

4. Real-time hazard analysis

,L D EDP IM M R

E L l f l d g f d edp f edp im f im g dLdDd EDPd IM

Exposure Vunerability Hazard

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


How to set the alarm threshold: a simple example

Let’s consider a simple school class equipped with a ringer and suppose that the students are trained to shelter under the desks when the alarm is issued.

7 m7 m6

m6

m

Desk

Lighting36 m2

What Causes Loss:

1.Structural collapse (causes death)2.Non-structural collapse (causes injury and eventually death)3.False alarm (cause panic and down-time loss)

Security action1.Sheltering below desks may reduce losses in the case of non-structural collapse

Expect

ed L

oss

[€

]

[s]

No alarm

Alarm

Optimal Alarm threshold

τ̂

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


M=6.6, Site at 131km far from the network

Naples Estimation of Magnitude

Estimation of Distance

Estimation of PGA at the site

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


Uncertainty Analysis

Uncertainty due to the attenuation law

Coeffi

cien

t of V

aria

tion

of th

e Es

timati

on

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


Uncertainty Analysis Findings

• Uncertainty on distance is confirmed to be negligible in EEW calculations;

• Also uncertainty on magnitude, altough larger than distance, is small if compared to that of the attenuation law;

• It is important to compute the hazard analysis including the uncertainty of the attenuation, which is the most important (this allows to copute missed and false alarm rates);

• To issue the alarm it is important to assess whether the level of information it is stable. E.g., triggering more station does not change the decision of alarming or not.

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


SABETTA E PUGLI ESE M=6 E R=120 km

0

0,1

0,2

0,3

0,4

0,5

0,6

0,7

0,8

0,9

1

0 0,1 0,2 0,3 0,4 0,5 0,6 0,7

PGA [m/s2]

P[PG

A>P

GA

c]

Decisional rule

[ ]C CAlarm if P PGA PGA P

PGAc

Pc

ALARM ! Because the probability that PGA exceeds the limit value is too high

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


:

:

True C

True C

MA no Alarm PGA PGA

FA Alarm PGA PGA

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


•Lead-time maps, defined on a 2700 nodes grid covering the whole Campanian territory, have been computed as a function of the number of stations triggered.

•Events occurring in the area covered by the ISNet infrastructure have been consideredred (hypocentral depths in Campania are small ranging from ~ 4 to ~ 12 km).

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


DESIGN MAP

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


Lead-time maps for the case-study region can be superimposed to real-time risk reduction actions for specific structural systems. These security

measures can be classified according to the time required to be carried out.

Info

rmati

on-d

epen

dent

lead

tim

e m

aps

for e

arth

quak

e ea

rly

war

ning

in th

e Ca

map

ania

regi


To the end of the project

• Design of a semi-active control system applied to a structure;

• The system may be tuned in real-time on the basis of a prediction of the response spectrum;

• Simulation of the system and risk assessment;

• If the EEWS lead to a reduction of the annual risk the EEWS is effective on the earhquake engineering perspective.

2nd year SAFER Project meeting. Armada Hotel, Istanbul, Turkey 25-27 June, 2008. Information-dependent lead time maps for earthquake early warning in the.

Documents

earthquake early warning

camapania region

alarm threshold

site slide

estimation slide

time loss security action

level of information

high slide