SOME DECLENSIONS OF SEISMIC RISK ANALYSIS FOR PERFORMANCE BASED ANALYSIS FOR PERFORMANCE‐BASED E ARTHQUAKE ENGINEERING Iunio Iervolino Iunio Iervolino Dipartimento di Ingegneria Strutturale, Ui iàd li S di di N liF d i II Universit àdegli Studi di Napoli Federico II
65
Embed
SOME DECLENSIONS OF EISMIC ISK ANALYSIS FOR … · SOME DECLENSIONS OF SEISMIC RISK ANALYSIS FOR PERFORMANCE‐BASED EARTHQUAKE ENGINEERING Iunio Iervolino Dipartimento di Ingegneria
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
SOME DECLENSIONS OF SEISMIC RISKANALYSIS FOR PERFORMANCE BASEDANALYSIS FOR PERFORMANCE‐BASED
EARTHQUAKE ENGINEERING
Iunio IervolinoIunio Iervolino
Dipartimento di Ingegneria Strutturale, U i i à d li S di di N li F d i IIUniversità degli Studi di Napoli Federico II
Topics1. REXEL: Computer aided code‐based record selection
Topicso pu e a ded code based eco d se ec o
Disaggregation MapsConditional hazard analysis for secondary IMs
2. Early warning as Real‐Time Performance‐Based Earthquake Engineering (RTPBEE);Real‐Time Probabilistic seismic hazard analysisReal Time Probabilistic seismic hazard analysisExpected‐Loss approach to early warning
3. Near‐Source probabilistic seismic hazard analysisDirectivity evidences in near source L’Aquila earthquake ground motion
4. Seismic risk analysis of classes of buildingsElastic period of RC building classes
An investigation about Eurocode 8 provisions for record g pselection
The set of accelerograms, regardless they are natural, artificial or simulated h ld h h f ll i i ishould match the following criteria:
- a minimum of 3 accelerograms should be used (7 to consider the mean effects on the structure);mean effects on the structure);
- in the range of periods between 0,2T1 and 2T1, the average spectrum cannot underestimate more than 10% elastic response spectrum from the codethe code.
Iervolino I., Maddaloni G., Cosenza E. (2008a). Eurocode 8 compliant real record sets for seismic analysis of structures. Journal of Earthquake Engineering, 12(1):54‐90.Iervolino I., Maddaloni G., Cosenza E. (2009a). A note on selection of time‐histories for seismic analysis of bridges in Eurocode 8. Journal of Earthquake Engineering. (in press)
What we found?it is almost unfeasible for practitioners to search in large databases to find i f l d ( ll l i ) h
What we found?
suites of seven real records (eventually multi‐component) whose average matches closely the design spectral shape without a specific software tool;
the spectra based on seismic zonation may be too severe in a way that do not exist suites of unscaled records whose average has such spectral shape;
it is not easy to control the variability (very large) of individual spectra in a combination, and this, in the fortunate case combinations are found, may impair the confidence in the estimation of the seismic performance usingimpair the confidence in the estimation of the seismic performance using such set;
the requirement of selecting records consistent with the earthquake eventsthe requirement of selecting records consistent with the earthquake events dominating the hazard at the site (i.e., the design earthquakes) requires PSHA/disaggregation data and skills seldom available to practitioners;
Iervolino I., Maddaloni G., Cosenza E. (2008). Eurocode 8 compliant real record sets for seismic analysis of structures. Journal of Earthquake Engineering, 12(1):54‐90.Iervolino I., Maddaloni G., Cosenza E. (2009). A note on selection of time‐histories for seismic analysis of bridges in Eurocode 8. Journal of Earthquake Engineering. (in press)
Seismic action in the new Italian codeSeismic action in the new Italian codeThe building code links the seismic design actions on structures directly to the PSHA results. In fact, the Istituto Nazionale di Geofisica e Vulcanologia(INGV) evaluated probabilistic seismic hazard for each node of a regular grid having 5km spacing and covering the whole Italian territory with over 103g p g g ynodes.
This resulted in hazard curves in terms of PGA (also disaggregation is provided for PGA) and spectral acceleration, Sa(T), for 10 different periods from 0.1 to 2 seconds Hazard curves are lumped in 9 probabilities of exceedance in 50 yearsseconds. Hazard curves are lumped in 9 probabilities of exceedance in 50 years (from 2% to 81%). All data can be accessed at http://esse1‐gis.mi.ingv.it.
NIBC acknowledges these data defining design spectra completely sitedependent, which, although given with standard functional forms, practicallycoincide with uniform hazard spectra (UHS) on rock for the site in question.coincide with uniform hazard spectra (UHS) on rock for the site in question.
Provisions of Italian code about recordProvisions of Italian code about record selection
As for EC8, the signals can be artificial, simulated and natural accelerograms. Three is still the minimum number of records to use, and seven allows the consideration of the mean effects on the structure as design value (rather than the maxima).
For the artificial records also the main condition the set of records should satisfy is the same as per EC8.
Real records or accelerograms generated trough a physical simulation of earthquake process, may be used, provided that, again as in EC8, the samples used are adequately qualified with regard to the seismogenetic features of the source and the soil conditions appropriate to the site. Selected real records have to be scaled to approximate somehow the elastic response spectrum in a range of period of interest for the structure.
Computer aided code‐based record selectionComputer aided code based record selection
Tolerances
Definition ofthe return O tithe returnperiod ofinterest
Options
Type ofsearchsearch
Records among to search amongIervolino I., Galasso C., Cosenza E. (2009c) REXEL 2.31 (beta) e la selezione normativa dell’input sismico per l’analisi dinamica non linearedelle strutture. Proc. of XIII Convegno ANIDIS, Bologna, Italy. (in Italian)
Convertito V., Iervolino I., Herrero A. (2009) The importance of mapping the design earthquake: insights for southern Italy. Bulletin of the Seismological Society of America. (in press)
Modal values from disaggregationModal values from disaggregation
PGA Sa(T=1s)
Convertito V., Iervolino I., Herrero A. (2009) The importance of mapping the design earthquake: insights for southern Italy. Bulletin of the Seismological Society of America. (in press)
Cosenza and Manfredi IM: Arias intensitydivided by PGA timesdivided by PGA timesPGV which has shown a good correlation with
li lcyclic structuralresponse.
t2( )
Eta t dt∫
0DI PGA PGV= ⋅
∫
Iervolino I., Manfredi G., Cosenza E. Ground motion duration effects on nonlinear seismic response. Earthquake Engineering and StructuralDynamics, 35:21–38, January 2006.
The Ignalina Nuclear Power PlantThe Ignalina Nuclear Power Plant
I th I li NPP th EEWS iIn the Ignalina NPP the EEWS is made up of a number of stations circling around the plant (Wieland et al 2000)et al. 2000).
Each station is made up of three sensors triggering at 0.025g. A 2‐
30km
out‐of‐3 logic is used to determine if a real seismic event has occurred.
Then an alarm is generated in the 30kmThen an alarm is generated in the reactor control room. Two seconds are needed to reduce the reactor capacity and prevent a meltdown p y pduring a severe accident. The false alarm issue is reduced by the
redundancy of measurement at the same geographic location.g g p
The Tohoku Shinkanzen is furnished with an EEWS which stops the trains if a
Morioka
certain acceleration treshold is exceeded at any of the accelerometers controlling aspecific segment of the track; this is to protect the trains from derailments.
The system was causing several delays due to a severe false alarm issue.
Tokio
Papadimitriou & Veneziano (1995) proposed a Tokioprobabilistic calibration of three different decisional rules strongly reducing the delay rate.
Regional Earthquake Early Warning Systems andRegional Earthquake Early Warning Systems and the ISNet Irpinia Seismic Network (Italy)
0 25 50 km0 25 50 km
Avellino
BeneventoCaserta
Napoli
SalernoPotenza
1981 – 2002 SeismicityINGV Seismic Catalogue
Earthquake M>3Earthquake M<31980 Earthquake Ms 6.9Seismogenetic faults of 1980 EarthquakeCurrent Seismic NetworkSeismic network under construction (2006)
Commonly used to give distributed estimates of the ground motion right after the
Main CityUrbanized area
Commonly used to give distributed estimates of the ground motion right after the event: SHAKEMAPS.
RTS: Rapid estimation of event magnitudeRTS: Rapid estimation of event magnitude
MT
Seismologists (i.e. Allen & Kanamori, 2003) claim it is possible to estimate the magnitude from the predominant period (τ) of the first 4 sec of the P‐wave velocity recording
RTS: Rapid estimation of event locationRTS: Rapid estimation of event location
EpicenterEpicenter
TriggeredTriggered Stations
Other seismologists (i.e. Zollo et al., 2007) claim it is possible to locate the hypocenter with negligible uncertainty within 4 sec from the event origin time
( ) [ ] ( ) ( )1 2 1 2| , ,..., | , ,...,| | , | |M R s s sf PGA f PGA m r f m f r s dr dm
ν ντ τ ττ = τ∫ ∫( ) [ ] ( ) ( )1 2 1 2| , ,..., | , ,...,M R s s s
M Rν ντ τ τ∫ ∫
PDF of magnitude Distribution of PGA at Negligible uncertainty
conditional on the measures of the seismic instruments
the site conditional on the measures of the seismic instruments
uncertainty
Iervolino I., Convertito V., Giorgio M., Manfredi G., Zollo A. (2006) Real time risk analysis for hybrid earthquake early warning systems. Journal of Earthquake Engineering, 10(6): 867–885. Convertito V., Iervolino I., Manfredi G., Zollo A. (2008) Prediction of response spectra via real‐time earthquake measurements. Soil DynEarthquake Eng, 28:492–505.
Lead time maps for the case study region can beLead‐time maps for the case‐study region can be superimposed to real‐time risk reduction actions for specific
A school classrom equipped with an EEWS terminal: howq ppto set the alarm threshold
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.
R l Ti l tReal‐Time loss assessmentE t di th h d h it i ibl t d t i th t dExtending the hazard approach it is possible to determine the expectedlosses condiotioned to the measurements of the seismic network in the
case of alarming or not
[ ] ( )( ) ( ) ( ) ( )| , |L D EDP PGA
E L l f l d g f d edp f edp pga f pga dL dD dEDP dPGAτ τ τ= ∫ ∫ ∫ ∫
Expected pLoss
1. Loss probability depending on the alarmin de ision
2. Structural damage probability depending on b ildin ’s seismi
3. Seismic response probability depending
on ha ard
4. Real‐time hazard analysis
alarming decision on building’s seismic response
on hazard
Iervolino I., Giorgio M., Manfredi G. (2007a) Expected loss‐based alarm threshold set for earthquake early warning systems. Earthquake Engineering and Structural Dynamics 2007; 36:1151–1168.
T t f th l th t i t dj t t t PSHA* i d iTo account for the pulse threat appropriate adjustments to PSHA* are required; in fact pulses are not always observed in near‐source conditions where directivity
effects are expected, therefore:
Att tiMAF Sa>x AttenuationMAF Sa>x(hazard)
Distribution of Magnitude and DistanceDistribution of Distribution of Magnitude and DistanceDistribution of pulse predictors
Distribution ofpulse period
*Iervolino, I, and Cornell, CA (2007). Prediction of the Occurrence of Velocity Pulses in Near‐Source Ground Motions. Unpublished manuscript.*Tothong, P., Cornell, C.A., and Baker, J.W. (2007). Explicit directivity‐pulse inclusion in probabilistic seismic hazard analysis. Earthquake Spectra, 23, 867–891.
Directivity effects’ predictors for strike‐slip (SS) and dip‐slip (DS) events*:
ParameterR [km] Closest distance of the site to fault rupture d [km] Distance to the site measured along the ruptureφ [deg] Angle between the rupture and the siteW [k ] Wid h f h f l
ParameterR [km] Closest distance of the site to fault rupture s [km] Distance to the site measured along the ruptureθ [deg] Angle between the rupture and the site[k ] h f h f l W [km] Width of the fault
L [km] Length of the faultX=s/L Length ratioM MagnitudeXcos(θ) Somerville’s amplification parameter
*Somerville, P.G., Smith, N.F., Graves, R.W., and Abrahamson, N.A. (1997). Modification of empirical strong ground motion attenuation relations to include the amplitude and duration effects of rupture directivity, Seism. Res. Lett. 68 199–222.
Pulse identification is based on that of Baker 2007*.
• Baker (2007) developed a method based on Lucerne (landers, 1992)( ) pwavelets to assign a score, a real numberbetween 0 and 1, to each analyzed record andto determine the pulse period. The larger thescore determined the more likely the record is Original GMto show a pulse.
• Herein only those in the fault‐normalcomponent have been considered; in
g
particular those ground motions which have apulse score larger or equal to 0.85 have been,arbitrarily, counted as pulse‐type records.
Extracted Pulse
• Baker analyzed extensively the NGA databaseand he is the only researcher the authors areaware of who has looked systematically at allrecords in the database. Therefore, we knowhi h th l d l hi h th
Residual GMwhich are the pulses, and also which are thenon‐pulses.
* Baker, J.W. (2007). Quantitative classification of near‐fault ground motions using wavelet analysis, Bull. Seism. Soc. Am. 97 1486–1501.
Iervolino I., Cornell C.A. (2008) Probability of occurrence of velocity pulses in near‐source ground motions. Bulletin of the Seismological Society of America, 98(5): 2262‐2277.
Iervolino I., Cornell C.A. (2008) Probability of occurrence of velocity pulses in near‐source ground motions. Bulletin of the Seismological Society of America, 98(5): 2262‐2277.
Iervolino I., Cornell C.A. (2008) Probability of occurrence of velocity pulses in near‐source ground motions. Bulletin of the Seismological Society of America, 98(5): 2262‐2277.
Iervolino I., Cornell C.A. (2008) Probability of occurrence of velocity pulses in near‐source ground motions. Bulletin of the Seismological Society of America, 98(5): 2262‐2277.
ObjectiveObjectiveThe objective is to compute the expected fraction in a class of buildings expected toreach a conventional limit state if subjected to seismic hazard in a given time.
The classes considered are those of existing rectangular RC buildings (both pre‐code or seisimically designed) defined by a limited number of parameters as geometricalseisimically designed) defined by a limited number of parameters as geometricaldimensions, and number of storeys.
a z
L z
a z
L z
a za z
L z
ayLy
ax
ayLyayLy
axax
LxLxLxLx
Iervolino I., Manfredi G., Polese M., Verderame G.M., Fabbrocino G. (2007b). Seismic risk of R.C. building classes. Engineering Structures, 29:813–820.
FormulationThe computation of the expected number of failures within the class proceeds extendingThe computation of the expected number of failures within the class proceeds extendingthe failure probability approach for a structure‐specific problem. The final result is thefailure probability of the class which in its frequentistic interpretation provides theexpected fraction of failures.p
Failure probability Class‐Capacity Function
( ) ( ) ( )0P P Z X P C X D X⎡ ⎤ ⎡ ⎤≤ ≤( ) ( ) ( )0fP P Z X P C X D X⎡ ⎤ ⎡ ⎤= ≤ = ≤⎣ ⎦ ⎣ ⎦
Limit State Function depending on the vector of random variables varying ithi th l d fl ti l
Seismic Demand
within the class and reflecting also bulding‐to‐building variability
Cl l it (2)Class‐scale capacity (2)A set of structures is defined to capture the variability of the structural features withinth l ( di t th di t ib ti f th i bl ) Th l d b SPOthe class (according to the distribution of the variables). These cases are analyzed by SPOand then the capacity features are obtained interpolating the results of the set ofstructures specifically analyzed.
Verderame G.M., Iervolino I., Manfredi G. (2009). Elastic period of sub-standard reinforced concrete moment resisting framebuildings. Bulletin of Earthquake Engineering. (submitted May 2009)