Seismic Risk Mitigation in Greece

DELFT UNIVERSITY OF TECHNOLOGY FACULTY OF CIVIL ENGINEERING & GEOSCIENCES DEPARTMENT OF DESIGN & CONSTRUCTION SECTION OF DESIGN & CONSTRUCTION PROCESSES Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Master Thesis in Building Engineering Graduation Committee Dimitris Detsis MSc graduate Building Eng Prof.dr.ir. H.A.J. de Ridder Design & Construction Processes (DCP) Dr.ir. R.B. Jongejan Hydraulic Engineering (WBK) / Jongejan Risk Management Consulting Ir. S. Pasterkamp Structural and Building Engineering (SBE) Ir. F.A.M. Soons Design and Construction Processes (DCP) Delft, June 2010

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Delft University of Technology Faculty: Department: Section: Civil Engineering & Geosciences Design and Construction Design and Construction Processes Report: Master Thesis in Building Engineering- Final Report Subject: Seismic Risk Mitigation in Greece Title: Translation of Dutch Flood Risk Management Practices Status: Final Report Author: Dimitris Detsis, MSc graduate Building Eng st.nr. 1535862 Graduation Committee Prof.dr.ir. H.A.J. de Ridder Design & Construction Processes (DCP) Dr.ir. R.B. Jongejan Hydraulic Engineering (WBK) / Jongejan Risk Management Consulting Ir. S. Pasterkamp Structural and Building Engineering (SBE) Ir. F.A.M. Soons Design and Construction Processes (DCP) Date: 14.06.2010 Contact details Design and Construction Processes Section Visiting address Stevinweg 1 2628 CN Delft The Netherlands Postal address P.O. Box 5048 2600 GA Delft The Netherlands Contact person Mrs. S.M.C. Hagman Room: 3.40 T: (+31) (0)15 2784774 E: S.M.C.Hagman@tudelft.nl Author Postal address 28th Octovriou, 16 15235, Vrilissia Greece (Athens) T: (+30)2106842676 M: (+30)6974805179 E: dedimitris@gmail.com Until 02/07/2010 Delft Hippolytusbuurt, 39A, 2611HM, Delft, The Netherlands T: (+31)(0)15-7851015 M: (+31)(0)6-12986561 *a CD with the model is attached to both hard-copy and digital version of this report D isclaimer Clause The best available sources and information have been used in preparing this report and these sources and information has been interpreted exercising all reasonable skill and care. Nevertheless, neither author nor reviewers accept any liability, whether direct, indirect or consequential, arising out of the provision of information in this report. ii 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Preface This report is the outcome of a master thesis in building engineering. The subject deals with seismic risk mitigation in Greece, this thesis focuses on translation (=the act of converting) of opportunities from flood risk management practices used in The Netherlands. This graduation project (thesis) consists of two phases - research and developing. Findings and results of both phases are reported here, followed by general conclusions. The main outcome is a proposal, towards the Greek authorities for the application of innovative seismic risk estimators. The report is delivered one week before defence presentation, which is going to be held on June 22 nd, 2010. TU-Delft, 14 June 2010 Dimitris Detsis, Dipl. Ing. - MSc graduate What is of most important for an engineer is to stay aligned with society needs and wishes. Acknowledgements First, I would like to thank all graduation committee members, Prof. de Ridder, Dr. Jongejan, Ir. Pasterkamp, and Ir. Soons. Consults, comments, and discussions with them guided my research. I appreciate the contribution of NTUA professor Th. Tassios and NTUA lecturer Emm. Vougioukas. All the material provided by them during my field research was more than valuable. Finally, I would like to thank those who supported me in any way during the preparation of this dissertation. Special thanks to Natalie G, the main reason and inspiration of this work. Dimitris Detsis iii

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Summary Seismic risk in some regions of Greece has increased over the last decades. The reason lies in urban development in earthquake prone regions, combined with a lack of ability or interest to tackle known construction vulnerabilities of buildings. Despite the severity of risk, also confirmed by recent events, homeowners have proven unwilling to mitigate seismic risk, possibly because of their inability to pay and/or unwillingness to invest due to lack of information or awareness. As a result, significant part of the building stock remains unsafe, in comparison with the safety level of the current building code. A way to stimulate seismic risk mitigation is government intervention. Such intervention could consist of a safety plan (retrofit program, mandatory insurance, emergency planning etc.) implemented by government, subsidies, introduction of more stringent building codes, risk communication. Recent developments in Greece regarding seismic safety are mostly aimed at vulnerability and risk evaluation, the publishing of a technical building Code of Interventions, and mapping out a seismic safety plan for Greece. Meanwhile, experts ask for government intervention proposing organisational change and a distribution of roles / liabilities among different clusters. For every scheme of government program, risk estimation is vital to be able to set priorities and decide whether buildings, municipalities, or regions are safe enough. Besides economic risks, risks to life should also be considered. Instruments for quantifying fatality risks are however unavailable at present. A review of the cornerstones of Dutch flood risk management practices, especially in risk estimation and decision-making has shown that fatality risks are considered from a societal perspective and an individual one. The societal risk metric concerns the (exceedance) probabilities of larger numbers of fatalities; the individual risk metric concerns the probability of death of a person at a specific location. In the case of the Netherlands, due to the nature of the flood hazard and protection scheme (public flood defences), the government is strongly involved in flood risk mitigation. Despite differences between the protection schemes for large-scale floods (strengthening dikes rather than protecting buildings) and earthquakes (strengthening buildings), this project proposes the translation (=the act of converting) of aforementioned metrics to the case of seismic risk in Greece. To quantify those metrics for earthquakes requires knowledge of the probabilities of different hazard levels (peak ground accelerations), the extent of damage on buildings given the hazard, and the expected number of fatalities in case of damage given the extent of damage. In this study, societal risk (depicted by FN-curves) and individual risk levels are quantified using exceedance probability function of peak ground acceleration at the site under iv 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report consideration, as well as deterministic transfer functions for damage (vulnerability curves) and losses (mortality curves). Moreover, since economic losses of earthquakes can also be significant, societal economic risk (FL-curve) and individual economic risk are also proposed and quantified. Using recent research results about the vulnerability of buildings, inventory data for social economic characteristics and reasonable assumptions about missing information (like building size); risk can be estimated for existing and retrofitted building stocks of Greek municipalities. After sensitivity analysis of model parameters, two case studies are presented that show the use of the aforementioned risk metrics for different levels of government decision-making. One simulating top level (central government) decision making, setting priorities for retrofit between municipalities, and the second simulating medium level (local government) decision making, setting priorities for a retrofit program between different structural typologies of buildings. The case studies show that the risk metrics and the model to quantify them can be useful tools for deciding which municipality should absorb more resources, whether mitigation is urgent, which mitigation strategy is most efficient, and how alternative retrofit programs influence risk levels. Of course, the model is only a prototype further refinements are advised. There are important benefits from the implementation of the described methodology. Firstly, the decision maker only deals with probabilities and consequences, has a general overview thus he/she may distribute resources and time in a more (cost) effective way. Moreover, human life is distinguished from cost-benefit analysis (no monetization). Events with high numbers of fatalities, which can cause disruption to the whole of the country, as well as disproportional individual exposures, can be targeted directly. Finally, it gives the opportunity to monitor the progress of a safety plan, and is scalable for central and local administration. This study concludes by proposing the application of societal & individual risk metrics (for fatalities and economic loss) to support two levels (central and local) of government decision making concerning seismic risk mitigation in Greece. Furthermore, it provides a prototype model for the quantification of these metrics. Finally, this thesis proposes directions for further research, the most important being research about the costs of alternative retrofit programs, which is necessary for the debate about appropriate (efficient/feasible) societal & individual risk acceptance criteria. Dimitris Detsis v

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Περίληψη (summary in Greek) Τις τελευταίες δεκαετίες η σεισμική διακινδύνευση (ρίσκο) σε ορισμένες περιοχές της Ελλάδα έχει αυξηθεί. Αιτία είναι η συγκέντρωση δομικού πλούτου σε σεισμογενείς περιοχές σε συνδυασμό με την καθυστέρηση στην αντιμετώπιση γνωστών κατασκευαστικών τρωτοτήτων σε υπάρχοντα κτίρια. Παρά τη σοβαρότητα της διακινδύνευσης, που επιβεβαιώνεται και από πρόσφατους σεισμούς, οι ιδιοκτήτες έχουν αποδειχθεί απρόθυμοι να προχωρήσουν σε επενδύσεις μετριασμού της, πιθανόν λόγω της οικονομικής δυσκολίας και/ή της απροθυμίας λόγω έλλειψης ενημέρωσης ή ευαισθητοποίησης να τις χρηματοδοτήσουν. Ως εκ τούτου, σημαντικό μέρος των κτιριακού πλούτου παραμένει ανασφαλές, σε σύγκριση με το επίπεδο ασφάλειας του ισχύοντα αντισεισμικού κανονισμού. Ένας τρόπος για να καταστεί δυνατή η μείωση της σεισμικής διακινδύνευσης είναι η παρέμβαση της πολιτείας. Οι στρατηγικές αυτής της παρέμβασης μπορεί να είναι γενικευμένο σχέδιο ασφάλειας (πρόγραμμα ενίσχυσης κτιρίων, αναγκαστική ασφάλιση, σχέδιο έκτακτης ανάγκης κλπ) που θα εκτελείται από τη πολιτεία, χρηματοοικονομικά κίνητρα, θέσπιση νόμων & κανονισμών και ενημέρωση περί της διακινδύνευσης. Οποιαδήποτε είναι η στρατηγική της παρέμβαση της πολιτείας, η εκτίμηση της σεισμικής διακινδύνευσης είναι ζωτικής σημασίας ώστε να είναι δυνατόν να καθοριστούν προτεραιότητες και να αποφασίζεται πότε ένα κτίριο ή μια περιοχή είναι επαρκώς ασφαλής. Εκτός από οικονομική διακινδύνευση, η διακινδύνευση της ζωής πρέπει να ληφθεί υπόψη. Παρόλα αυτά εργαλεία για την ποσοτικοποίηση της διακινδύνευσης ανθρώπινων απωλειών (ρίσκο θανάτου) δεν είναι ακόμα διαθέσιμα. Μετά από επισκόπηση στα κύρια χαρακτηριστικά των πρακτικών διαχείρισης του κινδύνου πλημμύρας στην Ολλανδία, ειδικά στον τομέα εκτίμησης της διακινδύνευσης και λήψης αποφάσεων, κύρια συμπεράσματα είναι η εφαρμογή δεικτών διακινδύνευσης όπως κοινωνικό ρίσκο (αφορά πιθανότητα υπέρβασης ενός γεγονότος και τον αριθμό των θανάτων αυτού) και ατομικό ρίσκο (πιθανότητα θανάτου ενός ατόμου στην διάρκεια ενός έτος σε μια συγκεκριμένη θέση). Στην περίπτωση της Ολλανδίας, λόγω της φύσης του κινδύνου των πλημμυρών και την διάταξη των μέτρων προστασίας, η παρέμβαση της πολιτείας συμβαίνει ήδη σε μεγάλο βαθμό. Παρόλο που υπάρχουν διαφορές μεταξύ των μέτρων προστασίας για μεγάλης κλίμακας πλημμύρες (ενίσχυση φραγμάτων και όχι κτιρίων) και σεισμούς (ενισχύοντας κτίρια), αυτή η εργασία προτείνει την κατάλληλη μετατροπή των προαναφερθέντων δεικτών για την περίπτωση της σεισμικής διακινδύνευσης στην Ελλάδα. Η ποσοτικοποίηση αυτών των δεικτών για τους σεισμούς απαιτεί γνώση των πιθανοτήτων διαφορετικών επιπέδων κινδύνου (δράση/ μέγιστη εδαφική επιτάχυνση), το εύρος της βλάβης σε κτίρια με δεδομένη την δράση, και την αναμενόμενες ανθρώπινες απώλειες δεδομένης της βλάβης του κτιρίου. Στην παρούσα μελέτη, καμπύλες κοινωνικού ρίσκου (FN-καμπύλες) και επίπεδα ατομικού ρίσκου υπολογίζονται προσεγγιστικά χρησιμοποιώντας την συνάρτηση vi 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report υπέρβασης πιθανότητας της μέγιστης επιτάχυνσης εδάφους (PGA), σε μια τοποθεσία, και ντετερμινιστικές συναρτήσεις για το υπολογισμό της βλάβης (καμπύλες τρωτότητας) και των απωλειών (καμπύλες θνησιμότητας). Επιπλέον, δεδομένου ότι οικονομικές απώλειες των σεισμών είναι επίσης ζωτικής σημασίας, προτείνονται αντίστοιχες εκτιμήτριες κοινωνικού οικονομικού ρίσκου και ατομικού οικονομικού ρίσκου. Χρησιμοποιώντας πρόσφατα αποτελέσματα ερευνών για την τρωτότητα των κτιρίων, στοιχεία απογραφής των κοινωνικοοικονομικών χαρακτηριστικών και εύλογες παραδοχές όσον αφορά στοιχεία που λείπουν (όπως η επιφάνεια των κτιρίων) οι παραπάνω εκτιμήτριες μπορούν να υπολογιστούν για τον υφιστάμενο, αλλά και «ενισχυμένο» κτιριακό πλούτο των ελληνικών δήμων. Μετά από ανάλυση ευαισθησίας παραμέτρων του μοντέλου, δύο υποθέσεις εργασίας παρουσιάζονται που δείχνουν την χρήση των προαναφερθέντων δεικτών διακινδύνευσης για διαφορετικά επίπεδα λήψης αποφάσεων. Το πρώτο αφορά στο ανώτερο επίπεδο (κεντρική κυβέρνηση), θέτοντας προτεραιότητες για ενίσχυση μεταξύ διαφορετικών περιοχών (ΟΤΑ) και το δεύτερο αφόρα στο μεσαίο επίπεδο (τοπική αυτοδιοίκηση) θέτοντας προτεραιότητες για ενίσχυση μεταξύ διαφορετικών δομικών τύπων κτιρίων. Οι παραπάνω υποθέσεις εργασίας δείχνουν ότι οι εκτιμήτριες διακινδύνευσης και το μοντέλο υπολογισμό τους, αποτελούν χρήσιμα εργαλεία για αποφάσεις όπως, ποιοι δήμοι πρέπει να απορροφήσουν περισσότερους πόρους, εάν είναι επείγουσες οι επεμβάσεις σε κτίρια ή μπορούν να προγραμματιστούν για το μέλλον, ποία στρατηγική μετριασμού είναι πιο αποτελεσματική, ποια είναι η επιρροή ενός συγκεκριμένου προγράμματος ενίσχυσης κτιρίων. Φυσικά το υπολογιστικό μοντέλο αυτής της εργασίας είναι μόνο ένα πρότυπο και χρειάζεται βελτιώσεις. Υπάρχουν σημαντικά οφέλη από την εφαρμογή της μεθοδολογίας που περιγράφεται. Πρώτον, η λήψη της απόφασης ασχολείται μόνο με τις πιθανότητες των απωλειών και τις απώλειες, δίνεται μια γενική εικόνα και έτσι μπορεί να γίνει αποτελεσματικότερη διανομή διαθέσιμων πόρων και χρόνου. Επιπλέον, πλεονέκτημα είναι η διάκριση της ανθρώπινης ζωής από την ανάλυση κόστους / οφέλους (αποφυγή κοστολόγησης της ανθρώπινης ζωής). Δίδεται η δυνατότητα να εκτιμηθούν η πιθανότητα γεγονότων με υψηλό αριθμό θανάτων, τα οποία είναι ανεπιθύμητα και μπορεί να προκαλέσουν αναστάτωση σε ολόκληρη τη χώρα, και οι δυσανάλογες εκθέσεις ατόμων σε διακινδύνευση. Τέλος, δίνει την ευκαιρία για παρακολούθηση της προόδου ενός σχεδίου ασφάλειας, και είναι εφαρμόσιμη και στα δύο επίπεδα λήψης αποφάσεων (κεντρικό & τοπικό). Η μελέτη αυτή καταλήγει προτείνοντας την εφαρμογή των δεικτών κοινωνικού και ατομικού ρίσκου (για ανθρώπινες και οικονομικές απώλειες), για την υποστήριξη της πιθανής παρέμβασης της κυβέρνησης σε δύο επίπεδα (κεντρικά και τοπικά) για την άμβλυνση της σεισμικής διακινδύνευσης στην Ελλάδα. Επιπλέον, παρέχει ένα πρότυπο μοντέλο για την ποσοτικοποίηση αυτών των παραμέτρων. Τέλος, προτείνει κατευθύνσεις για περαιτέρω έρευνα, η πιο σημαντική είναι η μοντελοποίηση εκτίμησης του κόστους προγραμμάτων ενισχύσεων, ώστε στη συνέχεια να μπορεί να ξεκινήσει η συζήτηση για τον προσδιορισμό και θέσπιση εφικτών / αποδεκτών κριτηρίων αποδεκτού ρίσκου. Dimitris Detsis vii

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Brief Contents.1 Introduction PART I GENERIC APPROACH.2 Hazard and risk.3 Criteria for retrofitting existing structures PART II COMPARISON/TRANSLATION.4 Flood risk mitigation in the Netherlands.5 Developments about seismic risk mitigation.6 Conclusion of research phase PART III SEISMIC RISK ESTIMATION MODELING.7 Description of developed estimation model.8 Sensitivity analysis Mitigation effectiveness.9 Potential use of metrics in safety policy PART IV - CONCLUSIONS.10 Conclusions PART V APPENDICES A. Model I-O volume and calculations B. Sensitivity analysis C. Choice of hazard curve D. Building behaviour assessment methods E. Building typologies Vulnerability curves F. Description of Greek building stock G. Retrofit Strategies H. Field research I. Personal Scope and Principles J. Societal & Individual Seismic Risk Estimator Model (S.I.S.R.E.M.) (excel file) viii 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Contents Preface... iii Acknowledgements... iii Summary... iv Περίληψη (summary in Greek)... vi Brief Contents... viii Contents... ix Figures... xi Tables... xii Glossary... xii Abbreviations... xiii.1 Introduction 1.1 Problem Analysis... 1-1 1.2 Graduation project research field... 1-3 1.3 Methodology... 1-4 1.4 Sources / Literature... 1-5 1.5 Thesis report... 1-6 1.6 Thesis flow-chart... 1-7 1.7 References... 1-8 PART I GENERIC APPROACH.2 Hazard and risk 2.1 Natural hazard risk management... 2-1 2.2 Seismic risk in Greece... 2-2 2.3 Flood risk in the Netherlands... 2-3 2.4 Comparison... 2-4 2.5 Risk perception... 2-5 2.6 Conclusions... 2-6 2.7 References... 2-6.3 Criteria for retrofitting existing structures 3.1 Formulating the problem... 3-1 3.2 Analysis... 3-2 3.3 Performance based design... 3-3 3.4 Limit state design... 3-4 3.5 Reliability index for existing building... 3-5 3.6 Eurocode enactment and recent developments... 3-6 3.7 Conclusions... 3-7 3.8 References... 3-8 PART II COMPARISON/TRANSLATION.4 Flood risk mitigation in the Netherlands 4.1 First formulation of the economic problem... 4-1 4.2 From economic model to cost benefit analysis... 4-4 4.3 Transferring the approach from Dutch major hazard domain... 4-6 4.4 Estimation of individual & societal risk... 4-8 4.5 Societal & Individual risk in the frames of safety policy... 4-10 4.6 Conclusions... 4-11 4.7 References... 4-12 Dimitris Detsis ix

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering.5 Developments about seismic risk mitigation 5.1 Seismic risk... 5-1 5.2 Seismic hazard... 5-1 5.3 Seismic vulnerability... 5-5 5.4 Earthquake losses... 5-7 5.5 Mitigation strategies... 5-10 5.6 Retrofit strategies... 5-11 5.7 Conclusions... 5-13 5.8 References... 5-14.6 Conclusion of research phase 6.1 Stimulating seismic risk mitigation... 6-1 6.2 Position of defence line in flood protection... 6-4 6.3 Conclusion... 6-6 6.4 References... 6-6 PART III SEISMIC RISK ESTIMATION MODELING.7 Description of developed estimation model 7.1 Philosophy... 7-1 7.2 Modelling risk estimation... 7-3 7.3 Main elements of calculations... 7-6 7.4 Inventory data... 7-12 7.5 Example... 7-13 7.6 Output / Results... 7-14 7.7 Modelling mitigations... 7-16 7.8 Conclusions... 7-19 7.9 References... 7-20.8 Sensitivity analysis Mitigation effectiveness 8.1 Sensitivity analysis... 8-1 8.2 Effectiveness of mitigation strategies... 8-3 8.3 Conclusions & Recommendations... 8-6.9 Potential use of metrics in safety policy 9.1 Seismic risk management steps... 9-1 9.2 Case study 1 - Top level decision making... 9-3 9.3 Case study 2 - Medium level decision making... 9-5 9.4 Evaluation of the proposed model... 9-9 9.5 Possible improvements / recommendations for research... 9-10 9.6 Societal & Individual risk limits... 9-12 9.7 Conclusions... 9-14 PART IV - CONCLUSIONS.10 Conclusions 10.1 Main conclusions (answers to research questions)... 10-1 10.2 The proposal... 10-2 10.3 Recommendations for research... 10-2 10.4 References (ALL)... 10-3 PART V APPENDICES x 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Figures Figure 1-1 - existing buildings intervention projects complex environment... 1-3 Figure 1-2 thesis flow chart... 1-7 Figure 2-1 - shift to more integrative hazard management... 2-2 Figure 2-2 tectonic plates around Greece (USGC)... 2-2 Figure 2-3 earthquakes with magnitude over 4 of the last 100years (EMSC)... 2-2 Figure 2-4 - Blue areas are threatened by flooding (Unie van Waterschappen, 2008)... 2-4 Figure 2-5 - Delta - works (Unie van Waterschappen, 2008)... 2-4 Figure 3-1 target Building Performance levels (FEMA-356, 2000)... 3-3 Figure 3-2 failure space as a function of basic (initial & normalised) variables... 3-4 Figure 3-3 economic decision problem (Vrijling, et al., 1998)... 3-5 Figure 3-4 - optimization of the construction and repair cost (Vrouwenvelder, et al., 2008). 3-6 Figure 4-1 cost minimization problem (Eijgenraam, 2006)... 4-2 Figure 4-2 loss curve (van Dantzig, 1956)... 4-3 Figure 4-3 dike rings & safety standards in The Netherlands (Brinkhuis-Jak, et al., 2004). 4-4 Figure 4-4 individual risk & limits (Jongejan, et al., 2009)... 4-7 Figure 4-5 societal risk & limit for establishments (Jongejan, et al., 2009)... 4-7 Figure 4-6 FN-curve for dike rings and total country (Maaskant, et al., 2009)... 4-9 Figure 4-7 influence of mitigation (Jongejan, et al., 2009)... 4-10 Figure 5-1 diagram of seismic loss rough estimation for building stock in a region... 5-1 Figure 5-2 acceleration graph... 5-2 Figure 5-3 - European-Mediterranean Seismic Hazard Map (Giardini, et al., 2003)... 5-2 Figure 5-4 - Greek Seismic Hazard Map (EPPO, 2003)... 5-2 Figure 5-5 example of PSHA result (Solomos, et al., 2008)... 5-4 Figure 5-6 simplified vulnerability curve... 5-5 Figure 5-7 risk of experience catastrophic losses (FEMA-389, 2004)... 5-9 Figure 5-8 characteristic capacity curves for retrofit strategies (EPPO, 2001)... 5-11 Figure 6-1 - decision making scheme... 6-3 Figure 6-2 different concept of the two cases... 6-5 Figure 7-1 diagram of the calculation methodology... 7-5 Figure 7-2 hazard curves for 3 seismic zones of EAK2003... 7-6 Figure 7-3 vulnerability curves for each typology (fitted to 2 nd polynomial)... 7-8 Figure 7-4 annual exceedance probability of damage for a building of each typology... 7-9 Figure 7-5 mortality in different damage states... 7-10 Figure 7-6 repair cost function... 7-11 Figure 7-7 - societal Risk for all area & for each typology... 7-14 Figure 7-8 societal economic risk for all area & for each typology... 7-14 Figure 7-9 building performance levels in FEMA-389 and BSSC 2001 (FEMA-389, 2004). 7-17 Figure 7-10 building performance levels chosen for this study... 7-17 Figure 7-11 retrofit levels and vulnerability of existing typologies... 7-17 Figure 8-1 cost function selection... 8-3 Figure 8-2 influence of migration strategies on FN curve... 8-4 Figure 8-3 influence of migration strategies on FL-curve... 8-4 Figure 8-4 - influence of mitigation strategies in AEF... 8-5 Figure 8-5 influence of mitigation strategies in AEL... 8-5 Figure 9-1 decision making structure... 9-1 Figure 9-2 - individual risk & individual economic risk... 9-3 Figure 9-3 - graph with societal risk (FN curve) of municipalities... 9-4 Figure 9-4 - graph with societal economic risk (FL curve) of all municipalities... 9-4 Figure 9-5 individual risk & individual economic risk for existing and retrofitted stock... 9-6 Figure 9-6 - graph with societal risk (FN curve) of all typologies and total... 9-7 Figure 9-7 - graph with societal economic risk (FL curve) of all typologies and total... 9-7 Figure 9-8 - graph with societal risk (FN curve) for 3 retrofit programs... 9-8 Figure 9-9 - graph with societal economic risk (FL curve) for 3 retrofit programs... 9-8 Figure 9-10 proposal for societal risk limits... 9-13 Figure 9-11 - Athens existing & retrofitted stock compared with limits... 9-13 Dimitris Detsis xi

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Tables Table 1-1 - research questions... 1-4 Table 1-2 compact steps of this report... 1-6 Table 2-1 natural hazard definition (Asian Disaster Preparedness Centre (ADPC))... 2-1 Table 2-2 steps of hazard management... 2-1 Table 2-3 major recent natural hazards (Institute of Local Government, 2008)... 2-3 Table 2-4 comparison table... 2-4 Table 3-1 - objectives according to FEMA-356... 3-3 Table 3-2 reliability indexes for different consequences class... 3-6 Table 4-1 variables of the economic decision problem... 4-1 Table 4-2 sensitivity analysis on Van Dantzig model... 4-2 Table 4-3 cost and benefits of flood risk mitigation... 4-5 Table 5-1 examples of earthquake scenarios exceedance probability... 5-3 Table 5-2 assessment methodologies (Karabinis, 2003)... 5-6 Table 5-3 seismic risk mitigation strategies... 5-10 Table 5-4 Retrofit strategies for r/c buildings (in Greece)... 5-11 Table 5-5 costs & Benefits of building retrofit... 5-12 Table 6-1 risk management steps for a government intervention... 6-2 Table 7-1 main output of the proposal... 7-2 Table 7-2 main inputs... 7-4 Table 7-3 annual exceedance frequencies of PGA (*g=10m/s 2 )... 7-7 Table 7-4 main structural typologies according to (EPANTYK, 2007)... 7-7 Table 7-5 mortality in different damage stage... 7-10 Table 7-6 retrofit programs... 7-16 Table 8-1 - model parameters and mitigation strategies... 8-3 Table 9-1 characteristic of cities... 9-3 Table 9-2 inventory of the four cities... 9-3 Table 9-3 societal risk limits... 9-13 Glossary translation in this project the act of converting μετάφραση risk probability of an event multiplied by its consequences διακινδύνευση ή ρίσκο risk metric a measure for risk; a means of deriving a δείκτης διακινδύνευσης quantitative measurement or approximation for risk (otherwise qualitative phenomena) [wiktionary.org] risk estimator a function or model that estimate certain risk metrics εκτιμήτρια διακινδύνευσης risk acceptance criteria κριτήρια αποδοχής ρίσκου societal risk criterion / metric κριτήριο/δείκτης κοινωνικού ρίσκου individual risk criterion / metric κριτήριο / δείκτης ατομικού ρίσκου seismic hazard σεισμικός κίνδυνος seismic vulnerability σεισμική τρωτότητα criteria that are used to express a risk level that is considered tolerable for the activity in question relates the exceedance probability of an event to the potential number of fatalities probability of death of an average, unprotected person that is constantly present at given location the potential for damaging effects caused by earthquakes [USGS, Putting Down Roots, 2007, Glossary] potential damage degree (0%-100%) of building resulting from a potentially damaging phenomenon xii 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Abbreviations A building area m 2 A build average building area in the examined region m 2 A col collapsed area m 2 AEF annual expected fatalities people AEL annual expected direct economic loss A floor average floor area in the examined region m 2 a g or PGA peak ground acceleration (*g) m/s 2 a o peak ground acceleration with exceedance probability 10% in 50 years (*g) m/s 2 C costs D damage degree % % E(l) expected losses F annual exceedance frequency (annual exceedance probability) 1/year F D mortality - F EV fraction of people evacuated - F EX fraction of people exposed - FL-curve annual exceedance frequency L loss representation of societal risk - FN-curve annual exceedance frequency N fatalities representation of societal risk - H hazard - h flood height m H future average height of dikes in the polder m H 0 existing average height of dikes in the polder m I investment (for floods construction costs of heightening) K S casualties (actually equals to N) people L or l direct economic loss or loss (in general) M 2,M 3,M 4,M 5 mortality function coefficient (see paragraph 7.3.3) - N or n fatalities people N P number of people in the dike people O area occupancy rate of the examined area (m 2 /people) m 2 /people p(n) probability of n fatalities occurs - P(N>n) probability of number of fatalities (N) to exceed n 1/year p f probability of failure - PV present value operator Q total cost RC reference rebuild cost S total damage in case of failure ( ) T L return period years T R mean or average return period years V vulnerability - V value of the goods in the polder w annual rate of exceedance 1/year X dike heightening m β reliability index (chapter 3) or voluntariness (chapter 9) Dimitris Detsis xiii

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Risico The deadliest flood in The Netherlands in 20 th century 1800 fatalities, 01/02/1953, North Sea storm Ρ ίσκο The deadliest earthquake in Greece in 20th century 470 fatalities, 12/08/1953, in Cephalonia island (Western Greece) xiv 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report.1 Introduction This introductory chapter firstly discusses background of the thesis, formulates the problem, and sets the objectives. It also presents research methodology, main & sub research questions and outlines the result. 1.1 Problem Analysis 1.1.1 Seismic risk for Greece Earthquake is a major hazard for many areas throughout the world; Greece is also facing this risk. Earthquakes visit the whole area of the country and often cause casualties and damages. Through history, builders were always aware of this problem; they were always trying to use technology to shield constructions against this hazard. However, deaths and damages after earthquakes in the 20 th century were always causing new developments in seismic risk mitigation. Ten years ago, the costliest earthquake in the history of Greece hit Athens, near 3 billion economic costs (Papazachos in Eleftherotypia, 2009). It was one more milestone for action from Greek society. Technical Chamber of Greece (TCG) president, Yiannis Alavanos, states that an average of 600 million is spent for seismic rehabilitation and post earthquake measures. Thus, he believes that it would be better if money were spent on seismic risk mitigation. (TO VIMA, 2009) Recent statistical research from sub organization of the TCG dealing with the National Policy for Strengthening Existing Structures (ΕPΑΝΤΥΚ) shows that from 4 million buildings (TO VIMA, 2009): 80% are built before 1985 (significant code change) and they are possibly vulnerable, strengthening cost can vary from 5% - 50% of their value Evaluation research about expected losses from earthquakes concludes that (Vlachos,1999): expected losses have been increased due to concentration of residential wealth in urban centres buildings with critical, obvious and known construction vulnerabilities are functioning regularly every effort of generalized intervention for the increase of seismic safety of existing building stock would required high costs Finally, according to Prof. Th. P. Tasios, president of above organization, the seismic risk mitigation is backed by the following two reasons (EPANTYK, 2007): demand for Dimitris Detsis 1-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering conserving of post earthquake losses (casualties, wealth etc) constitutional principle of equality in life expectancy and safety of property 1.1.2 Seismic risk mitigation in complex environment Besides high costs, another reason which obstructs seismic risk mitigation, is the complex environment. Concerning dwellings, more than 80% of the wealth is owned by the inhabitants. First idea would be that the owners should pay for the retrofit of their building. Unfortunately, for the majority of them it would be difficult to pay for such a project even if they are willing to mitigate seismic risk, while others will refuse to pay for extra costs for intervention because they are willing to accept level of seismic risk. The problem enlarges because of multi ownership in most of the dwelling buildings in large cities. Different willingness to pay and/or different perception of life safety value or financial incapability of the owner is usually the case. On the other hand, contractors / constructors are lacking experiences for this kind of projects regarding not only executing the technical parts but also coping with the administration issues. Researchers, although they have succeeded in providing with the new technology for the assessment of existing buildings, must now distribute it among the engineers and advisors in order to assimilate the new methodology. Engineers/ advisors should stand between contractors and owners to ensure the smooth running of those projects. Nevertheless, the most crucial role is governance and law. Government intervention could become very critical. State authorities can provide framework for the strategic design against earthquake. Experts should be asked to back a systematic policy for earthquake mitigation. Every one of the above stakeholders has different reaction to the social economic rehabilitation costs and different way to assess life safety value (see Figure 1-1). 1.1.3 Conclusion - Problem formulation Seismic risk in some regions of Greece has increased over the last decades. The reason lies in urban development in earthquake prone regions, combined with delay, lack of ability or interest to tackle known construction vulnerabilities of buildings. Despite the severity of risk, also confirmed by recent events, homeowners have proven unwilling to mitigate seismic risk, possibly because of their inability to pay and/or unwillingness to invest due to lack of information or awareness. As a result, significant part of the building stock remains unsafe, in comparison with the safety level of the current building code. 1-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Figure 1-1 - existing buildings intervention projects complex environment After long-range research in Greece about risk mitigation, experts ask for government intervention, proposing organisational change and a distribution of roles / liabilities among different clusters. 1.2 Graduation project research field 1.2.1 Project objective Considering the above conclusion, and within the frame of such a complex environment, this research project aims to give a proposal for instruments to support whatever government intervention. This proposal will be inspired by methods applied in the Dutch flood safety policy, after researching in cornerstones of flood risk management practices in The Netherlands. The focus will be more in risk estimation methodologies and decision-making instruments. 1.2.2 Research questions Main research question of this thesis is: How innovative & integrative risk metrics can be used in seismic risk estimation? Dimitris Detsis 1-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering While sub-questions that guide to answering main question follow (Table 1-1): I. Related to hazard management in general 1. How societies respond to hazards? 2. Which is the strategy for old structures strengthening? II. Related to comparison 3. How flood risk is estimated / mitigated in The Netherlands the last 50 years? 4. What are the developments in Greece about seismic risk estimation / mitigation? 5. Which elements of flood risk management are suitable for seismic risk management? III. Related to estimation modelling 6. How innovative & integrative risk metrics can be used in seismic risk estimation? Table 1-1 - research questions 1.3 Methodology 1.3.1 Research phase Introductory, question 1 regards hazard risk management in general; the answer sets the scene for Seismic Risk (S.R.) Management in Greece and Flood Risk (F.R.) Management in The Netherlands, while discussing general practices of hazard risk management. Question 2 relates S.R. Management with buildings and strengthening decision option, looking into methodologies suitable to estimate risk for a single building. After that, parallel, a description of F.R. estimation and mitigation in The Netherlands and S.R. in Greece is investigated (questions 3-4). This helps to spot out the translation opportunities and answers question 5. 1.3.2 Development phase After that, to support conclusion of research phase and go one-step further, a risk estimation model is developed based on the integrative risk metrics (societal & individual risk) and elements regarding seismic risk discussed. A dynamic model has been worked out combining real data and assumptions. Development phase aims to prove the suitability of centralized decision making for seismic risk mitigation in Greece. Therefore, the development phase presents the methodology, runs a sensitivity analysis to show criticality of parameters and comments on the results. Finally, two fictitious case studies are examined. The results are used to test the suitability of risk metrics to make decisions. The first case study refers to the comparison between the 4 largest cities of Greece, while the second refers to the municipality of Athens. Indicative results for seismic risk of regions in Greece are presented here. The development phase aims to answer research question 6 the main research question. 1-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 1.3.3 Conclusions phase Conclusions of this project concern: Potential of the use of societal & individual risk criterion in seismic risk estimation Necessary inputs and outputs for risk estimation model Recommendations for i. further research on specific critical inputs ii. refinements in the model iii. next steps regarding seismic risk estimation This study ends by proposing the application of societal & individual risk metrics (for fatalities and economic loss) to support two levels (central and local) of government decision making concerning seismic risk mitigation in Greece. 1.4 Sources / Literature 1.4.1 Literature This graduation project is based on several bibliographic references like scientific papers, regulations, official guidelines, and reports. The most important areas are: Reports and results of the program for retrofitting existing buildings TCG Scientific papers in the field of seismic risk estimation and mitigation Regulations and guidelines regarding risk management of earthquake hazard (FEMA-US) Scientific papers about the roots of flood risk estimation and mitigation and also about latest developments Research papers discussing the assessment of existing buildings in general and regarding the switch from national codes to Eurocodes. 1.4.2 Sessions, interviews and meetings During field research in Athens, there was an opportunity for an interesting and guiding discussion with Emm. Vougioukas lecturer of NTUA, member of TCG and key member of the committee National Program for Seismic Retrofit of Existing Building (EPANTYK). He explained the efforts made nowadays for enabling a building retrofit program and gave directions on where to investigate. It was also very educative, although slightly out of the scope, the seminar in Athens about the new technical code regarding building retrofit (KANEPE). More close to research objective was the presentation of the EPANTYK regarding a GIS system, which includes data about buildings in Greece, and it is a useful tool for risk estimation and decision-making. Findings from these sessions can be found in appendices G & H. Dimitris Detsis 1-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Finally, all graduation committee members gave critical recommendations and advices during bilateral and intermediate committee meetings. 1.5 Thesis report Table 1-2 compacts through the main steps of the report: PART I GENERIC APPROACH PART II COMPARISON/TRANSLATION PART III SEISMIC RISK ESTIMATION MODELING PART IV PART IV APPENDICES Chapter 1 - Introduction sets the scene of the problem and concludes to problem formulation describes research objectives, questions, methodology and sources Chapter 2 Hazard and risk (question 1) introduces natural hazard risk management cornerstones initiates description Dutch Flood Risk and Greek Seismic Risk discusses the concept of risk perception and government intervention strategies Chapter 3 Criteria for retrofitting existing structures (question 2) analyses developed theories for assessment & strengthening existing structures two main methodologies are described Chapter 4 Flood risk mitigation in the Netherlands (question 3) analyses cornerstones of flood risk estimation methods The Netherlands focuses in risk estimation methodologies describes innovative criteria, like individual and societal risk metrics Chapter 5 - Developments about seismic risk mitigation (question 4) analyses main principles of seismic risk estimation focuses in the case of Greece and developments there Chapter 6 - Conclusion of research phase (question 5) concludes the research phase discuss different concept of risk mitigation in 2 hazards and 2 countries propose the use of individual and societal risk metrics for the case of Greece Chapter 7 - Description of developed estimation model (question 6) outlines general philosophy of applying proposed metrics presents the methodology discuss the input data used presents the way mitigations are modelled Chapter 8 - Sensitivity analysis Mitigation effectiveness (question 6) presents conclusions of sensitivity analysis indicates the most critical parameters of the estimation model directs for future research Chapter 9 - Potential use of metrics in safety policy (question 6) describes main seismic risk management steps and positions the metrics, examines two case studies deals with recommendations for improvements (discussion about limits) Chapter 10 - Conclusions outlines main conclusions condenses the proposal of this thesis discusses main recommendation for further research A. Model I-O volume and calculations B. Sensitivity analysis C. Choice of hazard curve D. Building behaviour assessment methods E. Building typologies Vulnerability curves F. Description of Greek building stock G. Retrofit Strategies H. Field research I. Personal Scope and Principles J. Societal & Individual Seismic Risk Estimator Model (S.I.S.R.E.M.) (excel file) Table 1-2 compact steps of this report 1-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 1.6 Thesis flow-chart Figure 1-2 shows thesis flow chart. Figure 1-2 thesis flow chart Dimitris Detsis 1-7

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 1.7 References [1.1] ANTYK, Theodorakis S. and Vougioukas E. Priority policy for seismic retrofit / Seismic Retrofit of Existing Building (ANTYK). - Athens : Technical Chamber of Greece, 2001. - GREEK [1.2] EPANTYK Pre-seismic Retrofit of existing buildings / ed. Buildings National Program for Seismic Retrofit of Existing: Technical Chamber of Greece (T.C.G.), 2007. - GREEK. [1.3] Vlachos I. Estimation of expected economic losses from earthquake to building wealth of the country and their consequences: TCG - Technical Chamber of Greece, 1999. - GREEK. [1.4] Greek National Statistical Authority Buildings Inventory [Online] // GNSA. - http://www.statistics.gr. [1.5] Eleftherotypia The richters that changed us - Athens : Tegopoulos, 2009 Daily Newspaper [1.6] TO VIMA Seismic vulnerable 8 out 10 buildings Athens 2009. Daily Newspaper 1-8 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report PART I GENERIC APPROACH.2 references, and risk management methodologies. Finally, it discusses the concept of risk perception and government intervention strategies Hazard and risk Chapter 2 introduces natural hazard risk management and deals with definitions, historical 2.1 Natural hazard risk management 2.1.1 Natural hazard definition Those elements of the physical environment, harmful to man and caused by forces extraneous to him (Burton, et al.) a natural hazard is a threat, a future source of danger, that has the potential to cause harm to: People - death, injury, disease and stress Human activity economic, educational etc Property - property damage, economic loss of Environment - loss fauna and flora, pollution, loss of amenities Some examples of hazards are: earthquakes, volcanic eruptions, cyclones, floods, landslides, and other such events Human activities can cause or aggravate the destructive effects of natural phenomena; they can also eliminate or reduce them. Table 2-1 natural hazard definition (Asian Disaster Preparedness Centre (ADPC)) 2.1.2 Steps of hazard risk management 1. Identification What are the probable hazards and consequences? system definition (boundaries & critical functions) 2. Assessment & estimation What would be the magnitude of hazard? (hazard) What would be the degree of consequences in elements at risk? (vulnerability) What are the probabilities of certain consequences? risk metrics 3. Evaluation What is the significance of estimated risk? risk acceptance 4. Mitigation (if not accepted) Reduce probabilities Reduce consequences Table 2-2 steps of hazard management What should be done? Safety plan strategies: isolate, strengthen, insure, response, recover Dimitris Detsis 2-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 2.1.3 Comment on hazard risk management practices Through the history of human societies with the support of general development, hazard risk management is becoming more integrative. In the beginning, reaction - planning protection came after the impacts of hazard, while later developed societies planned protection using estimation and evaluation tools. (Figure 2-1) Nowadays policy makers are asking for more integrated safety policy, in which hazard probabilities and consequences are mitigated in conjunction. (Jongejan, et al., 2009). That is why modelling tools are necessary to be developed in order to support not only prevention planning, but also its evaluation, and controlling / monitor of its execution. Figure 2-1 - shift to more integrative hazard management 2.2 Seismic risk in Greece Earthquakes are a historical threat for Greece due to its position close to the collision zone between the Eurasian and the African lithosphere plates. The Aegean and surrounding area is seismically the most active region in the whole Mediterranean and in the whole West Eurasia.(National Observatory of Athens, 2001) Figure 2-2 tectonic plates around Greece (USGC) Figure 2-3 earthquakes with magnitude over 4 of the last 100years (EMSC) 2-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Consequences of earthquakes are very significant. Earthquakes affect directly infrastructure and buildings, lifelines, human behaviour while cause important indirect losses. The magnitude of the consequences depends, besides the intensity of the event, on the vulnerability of the aforementioned systems. Finally, to estimate expected consequences of an earthquake, economical and social information must be known for the affected systems (real / social / emotional value). Disaster scenarios are examined to estimate seismic risk for a region. Efforts for seismic risk mitigation in Greece are continuous while they are based in three axes, construction of seismic proof buildings, process, and application of seismic safety plans and informing/educating the public (further analysis in chapter.5). The following table shows the major natural hazard in Greece in the last century. date type region fatalities economic loss 1953 earthquake Cephalonia - Zakinthos 455 100 mil $ 1977 floods Athens 25 30 mil $ 1978 earthquake Thessaloniki 45 160 mil $ 1981 earthquake Corinth Athens - Viotia 20 900 mil $ 1986 earthquake Kalamata 20 745 mil $ 1987 heat - wave all country 700? 1990 drought all country 0 1300 mil $ 1995 earthquake Kozani - Grevena 0 450 mil $ 1995 earthquake Achaea Fokida 26 300 mil $ 1997 floods Athens Larisa Patrai - Volos 9 160 mil $ 1999 earthquake Athens 143 3300 mil $ Table 2-3 major recent natural hazards (Institute of Local Government, 2008) 2.3 Flood risk in the Netherlands Over the history, floods have been crucial hazard for The Netherlands. On February 1, 1953, the last severe flood struck large part of the country, when a huge storm caused the collapse of several dikes in the southwest of the Netherlands. The disaster left behind 1800 fatalities and severe damages to the infrastructure - 47000 buildings were destroyed. The total economic loss is estimated at 1.5 to 2 billion guilders (van Dantzig, 1956). The Dutch government subsequently decided on a large-scale program of public works, called "Delta Works" to protect the country against future flooding. Besides that, the Dutch scientists and engineers were commissioned to protect the country by monitoring and adjusting this protection system. The Delta Works and their strengthening project is a continuous process with considerable costs for the Dutch society. Risk management is valuable, in order to evaluate flood risk, make decisions, prepare and execute a safety plan. Dimitris Detsis 2-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Figure 2-4 - Blue areas are threatened by flooding (Unie van Waterschappen, 2008) Figure 2-5 - Delta - works (Unie van Waterschappen, 2008) 2.4 Comparison Table 2-4 is initiating discussion about comparison between natural hazards in 2 countries. Flood-The Netherlands Earthquakes-Greece INDENTIFICATION Casualties (last 60 yrs) 3 major events/1800 fat 12 major events/1000 fat Loss 1953: 2 b guilders 1953:100m$, 1999:3 b ESTIMATION / MITIGATION cause storm surge ground motion hazard water levels + waves earthquake PGA vulnerability existing height & strength of dikes existing building stock resistance defence future height of dikes in the polder future performance of buildings investment dike strengthening / heightening building retrofitting unit of area polder county SAFETY PLAN forecast (magn.-place) possible non possible evacuation possibility low (sea floods), moderate (river) very low to zero organization forced centralized individual asset Table 2-4 comparison table 2-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 2.5 Risk perception Hazard Risk Management is crucial for human societies; the main scope of HRM must be to minimize/mitigate risks for human lives and their wealth. Hazard risk estimation is one key step. HR estimation is a complex endeavour because of the double nature of it (Vrijling, et al., 2006): technical: determine risk by measurement and calculation, using by principle the equation probability of an event multiplied by consequences, non-technical: risk perception, which is people s judgment on characterization & evaluation of hazards While technical risk estimation can be done with engineering research & reasoning, nontechnical spreads to the field of social sciences. A risk metric estimated with technical way could have different impact to different individuals or societies. In general, from a collective decision-making point of view, four major response alternatives are available for the management of large-scale environmental risks (Vlek, et al., 1992): let us wait and see; it won t be so bad as some predict ; technological optimism, counting on the timely availability of technical solutions; psychological optimism: don t worry, people will adapt and survive fairly well ; and serious concern and concerted action towards significant changes in social and organizational behaviour patterns Therefore, after risk estimation and safety plan proposals by engineers or other specialists, social decision makers have the final call. Intervention strategies may be grouped into six general policy strategies useful for changing social behaviours in connection with environmental protection (Vlek, et al., 1992): 1. provide physical alternatives or rearrangements, 2. lawful regulation and enforcement, 3. financial economic stimulation, 4. providing information and communication, 5. social modelling and support, and 6. institutional and organizational change Nowadays new approaches are being developed trying to equalize the level of safety to the novel idea of acceptable risk. Aspects of acceptable risk: individual, who decides to undertake an activity weighing the risks against the direct and indirect personal benefits society, considering if an activity is acceptable in terms of the risk-benefit trade-off for the total population Dimitris Detsis 2-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 2.6 Conclusions 1. Hazard risk mitigation is necessary for protection of human societies. After identification of hazard, hazard risk estimation follows in order to provide information for the creation of safety plan for hazard risk mitigation. 2. Earthquake and floods threaten Greece and The Netherlands respectively. Even if the nature of the hazard is very different, main cornerstones of hazard risk management do not differ much. 3. When dealing with risk concepts, one has to bear in mind the double nature of it technical and non-technical. Furthermore, individuals & societies show different behaviour to risk. Novel approaches introduce acceptable risk concept as basis to decision-making. Respectively the two perspectives of acceptable risk are individual and societal. Government intervention may be grouped in six different general policy strategies and should be based on the aforementioned acceptable risks. 4. This study will focus on the technical estimation, having in mind that the results are not definite rather than forming the guidelines for social decision-makers of government intervention. This project deals with the translation of Dutch risk estimation methodologies to Greek case of seismic mitigation. The technical part of the risk referring to engineering would be on the focus, and not the decision making process which refers to policy. Going further to policy transplantation is not an objective of a technical dissertation but of a broader social analysis. 2.7 References [2.1] Burton I, Kates R.W. and White G.F. The Environment as Hazard [Journal]. - New York : Oxford University Press. [2.2] Jongejan R. B., Jonkman S. N. and Maaskant B. The potential use of individual and societal risk criteria within the Dutch flood safety policy (part 1): basic principles. Reliability, Risk and Safety, ESREL / ed. R.Bris C.G. Soares, S.Martorell. - 2009. [2.3] van Dantzig D. Economic Decision Problems for Flood Prevention / Econometrica. - The Econometric Society, July 1956. - 23 : Vol. 24. - pp. 276-287. [2.4] Vlek C. and Keren G. Behavioural decision theory and environmental risk management: Assessment and resolution of four survival dilemmas : Acta Psychologica, 1992. - Vol. 80. - pp. 249-278. [2.5] Vrijling J.K., van Gelder P.H.A.J.M. and Ouwerkerk S.J. Criteria for acceptable risk in the Netherlands / Infrastructure Risk Management Processes: Natural, Accidental, and Deliberate Hazards / book auth. Taylor Craig E. and Vanmarcke Erik. : ASCE, 2006. [2.6] Vrijling J.K., van Hengel W., l and Houben R.J. Acceptable risk as a basis for design / Reliability Engineering and System Safety. - [s.l.] : Elsevier, January 1998. - 1 : Vol. 59. - pp. 141-150. [2.7] Unie van Waterschappen Climate Change and Dutch Water Manageme - the Netherlands: Unie van Waterschappen, 2008. [2.8] Institute of Local Government Liabilities & Roles of Local Governments in Civil Protection for Mitigation of Natural Hazards. Athens : Institute of Local Government, 2008. - GREEK. [2.9] National Observatory of Athens [website: http://www.noa.gr/]. - 2001. 2-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report.3 Criteria for retrofitting existing structures Before opening the discussion for translation of Dutch flood risk management practices, this chapter analyses generally developed theories and methodologies for assessment and strengthening existing structures. After problem formulation and analysis, methodologies are described. 3.1 Formulating the problem Earthquake risk is not the only reason to assess the vulnerability of infrastructure and buildings. Structural behaviour of existing structures, even in non-earthquake prone areas, need to be assessed, because of 1. deterioration of their strength due to ageing and 2. synchronization with recent safety standards of human societies, resulting from technological development, as expressed in technical codes and regulations Until now, there are few regulations or legislations, which guide the assessment and the retrofit of an existing building for the above reasons. It is only required to prove, and if necessary upgrade, the structural safety of a building, during a renovation project and only if there is a change on the loading condition due to change of function. Otherwise, if a renovation project does not change the loads acting on the structure, only a declaration for structural safety by a structural engineer is necessary (no analytical assessment). Thus, until this moment it is the owner s choice to asses and, possibly, upgrade the structure. However, as far as the case of Greece is concerned, according to statistical data, almost 30% of the buildings are over 50 years old, over the reference design period. Furthermore, there is a demand for increased structural safety of constructions, built either under no or the old building code, not only because of seismic hazard but also because of aforementioned reasons. Similarly, in The Netherlands during the last years, there is a policy for more green and sustainable buildings, including also the existing stock. It is reasonable for renovation projects, along with the environmental performance, the structural safety upgrading to be considered, since lifetime is renewed. The issue is important because, among others, safety levels influence both the financial feasibility of individual renovation projects and the decision whether a building should be retrofitted or demolished. The issue influences in general macro-economy of society since building wealth and building industry comprise a large part of the total gross domestic product. Dimitris Detsis 3-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Therefore, the problem is which the criteria for strengthening existing structures/buildings are. Which theories are available for assessment and how these criteria can be set? 3.2 Analysis Criteria for safety acceptance of existing structures should be based on present guidelines and methodologies. According to Diamantidis, et al. (2007) the acceptability limits (or targets) should account for: 1. probabilities of failure, 2. potential losses, 3. amount of investments necessary to improve safety Such targets have been developed for various industrial activities including new civil engineering structures. Setting new codes and regulations encrypts criteria created by failure statistics, calibration to present practice and cost benefit analysis on behalf of the public. However, acceptable limits for new structures are not necessarily the same as those for existing ones, because of influencing factors such as: lifetime of the structure, degree of available technical knowledge (codes, methods, etc), relative effort (costs) to control safety, time constraints (for engineering and eventual repair), consequences of potential failures, socio-economical and political preference On the other hand, decisions regarding future actions to gauge performance of a structure or improve its safety may be: monitoring/control application of load restrictions change of use repair strengthening replacement In order to choose from the above alternatives you need to compare either: 1. probability of failure associated to a certain performance 2. limit state criterion compared with the acceptability limits The limits are decided after comparing marginal cost of reliability and consequences of failure. It is useful to note that these targets might be different for different stakeholders like owners, tenants, developers and public because each one has different priorities or time horizons. Respectively to the two alternatives, there are two different practices to assess a structure safety: Performance-based design and Limit state design 3-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 3.3 Performance based design Performance based design is a relatively new framework in which a range of performance objectives ( damages states) are coupled for a given level of hazard, with associated exceedance probability. (Diamantidis, et al., 2007) The need for performance based design appeared when it was realized that major earthquakes may not cause fatalities but are ending to unacceptable economic consequences (1989 Loma Prieta, 1994 Northridge and also Athens 1999). This meant that limit state design can provide acceptable level of safety against collapse (fatalities) but it is not adequate to protect economic wealth. In the US, National Earthquake Hazard Reduction Program (NEHRP) has introduced this type of design for the assessment of behaviour of building against earthquakes. The figure below presents the mapping between exceedance probability and different performance target (FEMA-356, 2000). Figure 3-1 target Building Performance levels (FEMA-356, 2000) According to FEMA-356, each cell in the above matrix represents a discrete performance target (rehabilitation objective). FEMA-356 defines 3 rehabilitation objectives (Table 3-1). For example, basic safety requires the structure to prevent loss of life for hazard level with exceedance probability 10% in 50 years and prevent collapse for hazard level with exceedance probability 2% in 50 years. Objective cells Basic Safety k + p Enhanced k + p + any of a, e, i, b, f, j, or n Limited k alone, or p alone Limited c, g, d, h, l Table 3-1 - objectives according to FEMA-356 Performance based design such as the above give the opportunity for flexible decisions depending on the marginal cost to increase safety. This means that you choose either from different values for the mean return period in case of the maximum credible earthquake and / or from different confidence level that the structure will satisfy the specified performance criterion. Dimitris Detsis 3-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering In general, performance-based design is more suitable for natural or accidental hazards. It is related to the entire structure than to structural components, so it gives a better depiction of the whole structure, thus structural elements and their connections. The new code of interventions (KANEPE, 2009), developed by Earthquake Planning and Protection Organisation (EPPO) in Greece, is introducing this kind of design theory (see appendix H). 3.4 Limit state design Limit state design defines a physical state of the structure that can be related to important consequences such as loss of serviceability, unsafe occupancy, or structural collapse. Limit state design assures that the structure will have sufficient strength and ductility capacity to withstand loading related to different states (ultimate, serviceability) with an acceptable failure probability p f or associated reliability index, β. There are proposed targets of p f or β depending on whether the failure is global or local, brittle or ductile and the significance of the failure consequences. Limit state design is more related to elements than the structural system and it is suitable for state of the art design in comparison with performance-based design, which is mainly associated with hazards. (Diamantidis, et al., 2007) Reliability index is the quotient of the average value and the standard deviation of the reliability function. Graphically it is represented from the origin to the failure space, which is described by a linear reliability function.(cur, 2006) Figure 3-2 failure space as a function of basic (initial & normalised) variables Reliability index for structural design according to Eurocode is the target value β=3,8 (reference period 50 yrs and 100yrs for bridges), corresponding to p f =10-4. Cost benefit considerations can be used to define risk acceptance and consequently target failure probabilities for structural components. For existing structures, there is a willingness to accept lower probabilities because costs of achieving a higher reliability level are high. Based on this economic argument, the reliability index for existing structures may be reduced. 3-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 3.5 Reliability index for existing building In more detail, the parameters that influence target probability p f / reliability index, β for the assessment building are residual service life, occupancy, social criterion (preservation value), activity, consequences of failure and warning / recognition of approaching failure. (Diamantidis, et al., 2007) The approach for the calculation of reliability index is an economic decision problem (Vrijling, et al., 1998) (Figure 3-3) min( Q) = min( I( P) + PV( P S)) f f (3-1) Q :total cost ( ) Ι :investment ( ) PV :present value operator S :total damage in case of failure ( ) The amount of damage S is increased with value of human value: S = P N s+s total d f i pi Where N pi number of participants in activityi. Figure 3-3 economic decision problem (Vrijling, et al., 1998) In Figure 3-4 the left graph shows that the minimization of the summation of construction costs and expected damage ( P S ) leads to β=3,8. In the right graph, thick line figures the f existing situation with assessed reliability index β=2,5 and risk B. The construction cost line now has larger steepness. This is because, as mentioned, costs of achieving a higher reliability level are higher. The optimisation point leads to β=3,2 (lower than the new situation). Finally, in order to be economically feasible to repair point A, it must be under point B. Thus, in the case of the figure if β=2,5 it is not accepted that the building should be demolished. Dimitris Detsis 3-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Figure 3-4 - optimization of the construction and repair cost (Vrouwenvelder, et al., 2008) Recent research (Vrouwenvelder, et al., 2010) ends up to the following table regarding the values of β for existing structures, concerning wind dominant (w d ) and not dominant building (w n ). These values concern human life safety normative, which in the majority of the cases is the governing criterion. Consequences Minimum reference period β-new β-existing class for existing building w n w d w n w d 0 1 3,3 2,3 1,8 0,8 1 15 3,3 2,3 1,8 1,1 2 15 3,8 2,8 2,5 2,5 3 15 4,3 3,3 3,3 3,3 Table 3-2 reliability indexes for different consequences class Accepting lower reliability index means the use of lower γ values. Therefore, the factors for the loads are decreased to 0.10, 0.15, and 0.20 respectively to consequences classes CC1, CC2, and CC3. 3.6 Eurocode enactment and recent developments Eurocodes form a separate category of European standards. They are guidelines intended for use in building structures. These Eurocodes are comparable to the Dutch Technical Guidelines on Structures (TGB). In 2010, the use of Eurocodes will be obligatory for the countries of EU. However, there is no official technical document for the structural safety of existing buildings (EC 8 part 3 refers to seismic loads). For that reason new NEN 8700 Dutch is about to be developed. The main reason for the publishing of such a code is that, if new safety levels are applied to existing structures or renovation projects, costs are going to be increased, making these projects unfeasible. Thus, safety levels should be reduced, as explained above. 3-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 3.7 Conclusions 1. Structural behaviour of existing structures, even in non-earthquake prone areas, need to be assessed, because of deterioration of their strength due to ageing and synchronization with recent safety standards of human societies. The safety assessment of an existing structure differs from a new one, because in order to increase safety you have to spend a lot more, the remaining lifetime is less than the standard period of 50years and finally performance measurement can be made to reduce uncertainty. 2. There are two different practices to assess a structure safety: Performance-based design and Limit state design. In the first a range of performance objectives ( damages states) are coupled for a given level of hazard while the second assures that the structure will have sufficient strength and ductility capacity to withstand loading related to different states (ultimate, serviceability) with an acceptable failure probability p f or associated reliability index, β. 3. Performance based design is more suitable for hazard loads and it is related to the entire structure than to structural components, so it gives better depiction of the whole structure, including structural elements and their connections. On the other hand, limit state design is more related to elements than the structural system and it is suitable for state of the art design. 4. Safety standards for existing building can be lower than new ones. In the case of performance-based design, one can choose for lower performance target/objective. While in limit state design, load factors are accepted to be lower. 5. In The Netherlands, the switch from national codes to Eurocodes is done by checking the individual risk metrics. If the building is safe enough then cost benefit analysis will answer the question renovate or demolish. In the later calibration of reliability index β is proposed to consider increasing costs and reduced remaining lifetime. Dimitris Detsis 3-7

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 3.8 References [3.1] CUR Probability in Civil Engineering - Delft : TUDelft, 2006. - Vol. 1. - Lecture Notes (CT4130). [3.2] Diamantidis D. & Bazzurro P. Safety acceptance criteria for existing structures / Workshop on Risk Acceptance for existing structures. - Stanford University, USA, March 2007. [3.3] FEMA FEMA 356 - Prestandard and commentary for the seismic rehabilitation of buildings.- Washington DC : American Society Of Civil Engineers, 2000 [3.4] Priemus H. Unification of the European building market: Possible consequences for the Dutch construction industry / Housing and the Built Environment. - March 1991. - 1 : Vol. 6. - pp. 35-46. [3.5] Vrijling J.K., van Gelder P.H.A.J.M. and Ouwerkerk S.J. Criteria for acceptable risk in the Netherlands / Infrastructure Risk Management Processes: Natural, Accidental, and Deliberate Hazards / book auth. Taylor Craig E. and Vanmarcke Erik. ASCE, 2006. [3.6] Vrijling J.K., van Hengel W., l and Houben R.J. Acceptable risk as a basis for design / Reliability Engineering and System Safety. Elsevier, January 1998. - 1 : Vol. 59. - pp. 141-150. [3.7] Vrouwenvelder T. and Scholten N. Assessment criteria for Existing structures / Structural Engineering International / ed. Editorial SEI. - November 17, 2010. - Vol. 1. [3.8] Vrouwenvelder A.C.W.M. and Scholten N.P.M. Veiligheidsbeoordeling bestaande bouw - Achtergrondrapport bij NEN8700 / Bouw ; TNO. - Delft : TNO Bouw, 2008. - p. 39. - 2008-D-R0015/B. 3-8 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report PART II COMPARISON/TRANSLATION.4 Flood risk mitigation in the Netherlands Cornerstones of flood risk estimation methods implemented in The Netherlands are discussed in this chapter, since the first formulation of the economic problem after 1953 disaster until the recent developments, regarding the transfer of safety policy from man-made hazards. Cost benefit analysis and innovative criteria, like individual and societal risk metrics, as used in the case of flood risk estimation, could be useful instruments for seismic risk estimation. 4.1 First formulation of the economic problem 4.1.1 Economic decision problem According to van Dantzig (1956) for flood risk in The Netherlands three categories of mathematical problems can be distinguished: 1. Statistical problems, which are about statistically estimation of sea level heights, 2. Hydrodynamic problems, concerning the height of sea level that a storm can cause, 3. Economic decision problems Van Dantzig modelled the decision on increasing flood defences capacity. He proposed the following problem formulation: Taking into account the cost of dike-building, the material loses when a dike-break occurs, and the frequency distribution of different sea level, determine the optimal height of the dikes The problem now can be translated as the determination of X, dike heightening, in order to minimize the sum of construction costs and insurance costs: MIN( I( X ) + L( X )) di dx + dl dx = 0 (4-1) The variables to solve this problem are: h flood height (m) H 0 existing average height of dikes in the polder (m) H future average height of dikes in the polde r (m) X = H H, dike heightening to achieve H 0 (m) I = I( X ), constructio n costs of heightening ( ) V value of the goods in the polder( ) S = V (if h H ) loss( ) probability distribution that any height h will be exceeded at least once during a year ph ( ) ah ah ( ph () = ce = p e H 0 ) 0 (Wemelsfelder), where p = c e ah L t= 0 0 = ph ( ) V (1 + δ) t, sum which is invested at a rate of interest δ and must cover the expected values of future losses, ph ( ) V, each year (insurance problem) Table 4-1 variables of the economic decision problem 0 Dimitris Detsis 4-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Considering a relatively small interval of values X and I a linear function of X can be assumed. The minimization problem ( Figure 4-1) has the following solution: 1 p V a = ln δ (4-2), where k is the subsequent cost of heightening/ meter 0 X a k Figure 4-1 cost minimization problem (Eijgenraam, 2006) 4.1.2 Analysis To understand the physical meaning of equation 4-2, the influence of several parameters to the strengthening height is presented here. X increases if: p increase 0 V k increase decrease a decrease δ decrease Table 4-2 sensitivity analysis on Van Dantzig model It is also interesting, how the model encompasses technical (physical or statistical) and economical components. The first are expressed by constants such as and p o & α. Technical constants are subjected to epistemic uncertainty; they can be more accurately determined by further research and mathematical / statistical treatment of the data. In contrast, later components such as k (commercial related) and V (value related) include subjective uncertainty, which make it hard to estimate. For value -V - in risk (or loss), estimation is far more complicated since it includes consequential/indirect loss. Regarding it is an index number (an economic data figure reflecting price or quantity compared with a standard or base value problem) and it is by no means certain that an index can be defined with sufficient precision, so that the quantitative comparison of values is centuries. Consequently, economic data is difficult to be assessed accurately. δ 4-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 4.1.3 Points of interest The above decision methodology is the first cost benefit analysis approach to the flood risk mitigation problem. Van Dantzig discussed also in his proposal the following notable aspects. Evaluation or even monetization of human life and other ideal values can be a very difficult endeavour. This is because even the most detailed calculation cannot take into account the emotional factor. Furthermore, it must be considered that loss curve caused after a flood event is not linear (Figure 4-2) this means that beyond a point it becomes extremely high, while it is hard to estimate the probability of occurrence of such an event and the loss caused by it. Someone will be willing to spend a multiple amount of the one that would be lost in flood, if the flood can be prevented. However, the factor cannot be determined on mathematical, statistical, or economic grounds; it is a responsibility of authorities not of scientists to determine this factor. Figure 4-2 loss curve (van Dantzig, 1956) Van Dantzig (1956) and Van Dantzig & Kriens (1960) used economical optimization of the height of the flood defences along the Dutch coast resulting in a minimum safety level of main sea dikes of 1/125000 /yr which corresponds in safety standard exceedance probability P(H>h)=10-4 /yr. (Vrijling, et al., 2006) Closing, drawback of the van Dantzig model is that it is stationary. There are no time dependencies, it looks only to how much? but does not answer to when? or when again? Furthermore, investing as much that flood probability returns to an old level is inconsistent with economic growth. A dynamic cost minimization would include deterioration of the water system (e.g. climate change and subsidence of land), which increases the probability of floods and growth of population and wealth, which increases the expected losses. (Eijgenraam, 2006) Dimitris Detsis 4-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 4.2 From economic model to cost benefit analysis 4.2.1 Water Defence Act Until 1953 dikes were constructed to withstand the highest known water level, and the decision making philosophy was reactive. The mathematical optimization, described in the previous paragraph, can be considered a rough cost benefit analysis, based on which major investments to improve water defence were decided. The Delta Works were given a priority to protect the country against inundation. After that, the strengthening of the river dikes began. However, other parameters needed to be considered also, like influence in the environment. Dike s strengthening program resulted in loss of nature areas, landscape, and sites of cultural value. In 1993, the Dutch government and parliament agreed on a new approach in order to protect landscape and natural areas. The floods of 1993 and 1995 drew attention about risks of life, so the water defences were reinforced faster.(brinkhuis-jak, et al., 2004) Most of water defences were completed in 2001 and complied with Water Defence Act. Main characteristics of them are: 1. country is divided in 53 dike rings (Figure 4-3), 2. dike rings are areas protected against series of water defences like: Dikes, dunes, hydraulic structures, & high grounds 3. standards for these dike rings are based on probability of exceedance of a water level that flood defences should be able to safely withstand 4. design and safety evaluation are based on the above water level Figure 4-3 dike rings & safety standards in The Netherlands (Brinkhuis-Jak, et al., 2004) 4-4 14 June 2010

4.2.2 Recent developments of flood mitigation Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Besides dike strengthening, as a measure for flood risk reduction, more innovative ways of mitigating risks have been developed. One main development was the idea of giving rivers more room; this was combined with national policy of spatial planning. Another way was safety planning considering emergency retention areas; areas that can be inundated in a controlled way to prevent uncontrolled flooding of an area. Furthermore, evacuation planning and insurance of flood damage can reduce consequences, if a flood occurs. It is obvious that more options are available, which also might be more effective as well. Thus, it is necessary to quantify cost and benefits among different alternatives of mitigation measures. 4.2.3 Cost-benefit analysis The basic principle of cost benefit analysis (CBA) is that a project results in an increase of economic welfare, which is translated, the benefits of the project must exceed the costs. In flood risk management, the costs of measures for increasing the safety against flooding are compared with the decrease in expected flood damage. The main cost and benefits to be considered are summarized in the following table: Costs Benefits Mitigation Investment design & construction Reduction of possible damages maintenance management Potential of economic growth Table 4-3 cost and benefits of flood risk mitigation Main weakness of CBA is that it needs quantification of all costs and benefits in monetary terms. It is difficult in many cases to quantify indirect loss or economic growth, however if done correctly with clear and explicable steps, CBA is a useful and sound method. Thus, it can provide valuable indications for decision-making. What actually differentiates cost benefit analysis from van Dantzig optimization model is that the latter does not consider economic benefits of risk reduction in the area due to a flood protection project. 4.2.4 Application Main components for application cost benefit analysis for a flood reduction investment are: 1. flood probabilities 2. investment costs - construction and maintenance, and also indirect costs like loss of agricultural areas, disturbance during the work etc., and 3. damages caused by floods Dimitris Detsis 4-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Probabilities for the existing situation are already calculated for the dike rings (Technical Advisory Committee for Water Defences) while probabilities after mitigation measures can be modelled. For the estimation of flood damage, damage functions are used. Damage curves are relating flood height with inundation depth and use. However, damages include also intangible loses, which due to difficulty to be indentified and monetized, may be under-estimated. A generalized formula of CBA for a mitigation investment is written below (Brinkhuis-Jak, et al., 2004): I + I '( ln( P )) < ( P P ) L/ r' (4-3), where: 0 f f,0 f I0 + I'( ln( Pf )) = I : investment - related logarithmically linear with probability - ( ) L/ r ' P & f,0 P f : damages / losses discounted by r (discount rate) - ( ) : probability of initial and future situation Concluding, cost benefit analysis and economic optimization model leads to measures cost effectiveness of a mitigation strategy for the decision maker. The strategy that achieves the largest risk reduction with the smallest investments is the optimum solution. However, because of the difficulty to quantify accurately intangible costs of mitigation before or intangible damages after a flood event, it is recommended to be used in broader framework of decision making, in which other parameters / values like ecological, social and political could be considered. 4.3 Transferring the approach from Dutch major hazard domain 4.3.1 Integrated flood safety According to the methodology described in the previous chapters, minimizing the sum of investments in flood defence and the discounted expected value of future loss could only include various monetized intangible losses, like loss of life. Contrary to that integrated flood safety policy flood probabilities (other than exceedance probabilities of loading conditions) and flood consequences are mitigated together. The latest efforts and research in The Netherlands is about transferring this approach used in the Dutch Major Hazard domain to flood safety. The methodology used within the Dutch major hazard policy has the following cornerstones (Jongejan, et al., 2009): 1. quantitative risk analysis, 2. individual and societal risk and 3. quantitative acceptability criteria for evaluating levels of individual & societal risk 4-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Latest developments on flood policy in The Netherlands include the application of this methodology, in order to give a better insight to policy makers to flood probabilities and consequences in a more integrated way. This safety policy in its current form originated in the beginning of the 1980s, from the application on LPG, and then it became clear that the use of LPG would increase considerably quantitative assessment of risks and quantitative criteria for decisions on risk acceptability. (Vrijling, et al., 2006) 4.3.2 Basic risk metrics Basic risk metrics of this integrated methodology are individual risk and societal risk. Individual risk is defined as the probability of death of an average, unprotected person that is constantly present at given location. The limits to individual risk, given by External Safety Decree (2004), prevent disproportional individual exposures. Individual risk is represented with contours lines in the region examined (Figure 4-4). However, individual risk criterion alone cannot prevent the too frequent occurrence of multifatality accidents. These events have a considerable impact on the society and they are often considered unacceptable. That is why societal risk criterion is introduced, relating the exceedance probability of an event to the potential number of fatalities. Societal risk is graphically represented by an FN-curve that shows the exceedance probabilities of the potential numbers of fatalities on double log scale. The Dutch Major Hazard policy societal risk criterion is 10-3 /n 2 per installation per year. The following figures show an example of societal risk (Figure 4-5). Figure 4-4 individual risk & limits (Jongejan, et al., 2009) Figure 4-5 societal risk & limit for establishments (Jongejan, et al., 2009) Dimitris Detsis 4-7

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 4.4 Estimation of individual & societal risk 4.4.1 Quantitative flood risk analysis To quantify flood risk the next steps need to be followed (Jongejan, et al., 2009): 1. probability density functions of hydraulic conditions, 2. probability density functions of the variables that determine the load bearing capacity of flood defence, 3. fault tree models to analyze failure modes, 4. flood propagation models i. land-use data ii. loss functions 4.4.2 Mortality functions The number of fatalities for a particular flood scenario depends on (Jongejan, et al., 2009) 1. spatial distribution of flood characteristics 2. population densities of affected regions 3. possibilities for evacuation 4. probabilities of death of the affected individuals Mortality functions are used to estimate the probability of death at a given location for any given scenario. If mortality function is integrated within the population density of the affected region, then the overall probability density function is calculated: = i n P(, x y) p ( c(, x y)) p d d i i= 1 (4-4), where: Pd ( x, y ) : probability of death at location ( x, y) per year p ( c( x, y )) : probability of death in case of flood characteristic at d n p i i : number of flood scenarios : probability of scenario i per year c ( x, y ) scenario i Since extreme-river discharges can be forecasted days ahead, there is a potential for effective evacuation. The effectiveness of evacuation depends on: 1. Reliability of the forecast 2. Preparedness 3. Evacuation rate 4.4.3 Estimation of individual & societal risk By applying a simplified approach, individual & societal risk for dike rings in The Netherlands can be determined. The following two figures are needed (Maaskant, et al., 2009): 4-8 14 June 2010

1. probability of flood and 2. average number of fatalities in a flood This based on the following assumptions: Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report smallest probability goes with the largest number of fatalities etc, probability of flood equals the probability of at least one fatality and number of fatalities given flood is exponentially distributed, Results in the following formula P( N n) P e Ν = (4-5) f n Flood probabilities P f were estimated by a large ongoing project elaborated by Dutch government. While N, expected number of fatalities is determined by several variables and according to the following formula: N= N F (1 F ) F (4-6), where: N P F EX F EV F D P EX EV D :number of people in the dike :fraction of people exposed :fraction of people evacuated :mortality 4.4.4 Results Until now in The Netherlands, societal risk for several dike rings is calculated. In order to estimate it for the whole of the country, cumulation (dependencies between failures) has to be taken into account. The following figure is showing the FN curves for the 53 dike rings (thin/colour lines) and for the country (bold/red line), the later is estimated only by vertical adding curves of individual dike rings, meaning that floods in different dike rings are considered mutually exclusive events (Maaskant, et al., 2009). Figure 4-6 FN-curve for dike rings and total country (Maaskant, et al., 2009) Dimitris Detsis 4-9

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 4.4.5 Modelling mitigation It is interesting to examine the impact to FN curve after mitigation. The results are depending on the mitigation strategy. Main strategies for flood risk mitigation can be distinguished in two categories, those that aim to reduce flood probabilities (dike strengthening, beach nourishment, widening rivers etc) and those that aim to reduce flood consequences (flood proof construction, evacuation opportunities, safety zoning, splitting dike rings).the FN-curve is influenced differently by mitigation (Figure 4-7): 1. By reducing flood probabilities FN curve will shift downwards (first graph) 2. By reducing flood consequences FN curve will shift to the left (second graph) Figure 4-7 influence of mitigation (Jongejan, et al., 2009) 4.5 Societal & Individual risk in the frames of safety policy The main advantage of using societal and individual risk metrics in the frames of flood safety policy is that fatalities are considered separately from cost benefit analysis. This gives the opportunity to have an insight of risk for the population of different dike rings. Moreover, vagueness of monetization for human life is no longer present. In addition, using both societal and individual risk criteria gives a perspective for the safety of an individual and the society probabilities of large numbers of fatalities that can have a major impact on the country. The fact that the FN curve, as a representation of societal risk, gives also a view to the influence of different mitigation strategies, helps decision makers to assess the effectiveness and decide upon portfolio of investments in projects that manage risk. Additionally, limits can be set to guarantee minimal safety for both metrics. A relative example is the limit proposed by the Second Delta Committee (2008) for individual risk to be less than 10-6 for the polders. For the societal risk, lines on the FN graph can be limits. Further research is needed to check the feasibility of the proposed limits in relation with the measures that are necessary to be taken. (Jongejan, et al., 2009). Finally, individual and societal risk can initiate the discussion for new measures or control the process of an ongoing safety plan with mitigation projects. 4-10 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 4.6 Conclusions 1. Floods have been a crucial threat for The Netherlands; they are usually low probability with high impacts events. After the disaster of 1953, besides statistical and hydrodynamic problems, Van Dantzig approached flood safety as an investment problem. This model finds the optimal height of a dike considering the cost of building it, the material loses when a dike-break occurs, and the frequency distribution of different sea level. Even though flood probabilities were not estimated back then, van Dantzig noticed the difficulty to monetize the intangibles like human life, model loss function and assess willingness to pay for a safety plan. 2. Until 2001, while Delta Works were completed, more innovative ways of mitigating flood risk have been developed. Thus, it is necessary to quantify cost and benefits among different alternatives. Cost benefit analysis is used to give indication about their effectiveness. More information is now available, like damage functions, but intangible losses are still roughly monetized. CBA should be supplementary to broader decision making, in which other parameters / values including ecological, social, and political could be considered. 3. The latest research in The Netherlands is about transferring Major Hazard Safety Policy to the domain of flood protection. The approach used for man-made hazards is more integrated, meaning that probabilities (other than exceedance probabilities of loading conditions) and consequences are mitigated together. The cornerstones of this policy are quantitative risk analysis, individual and societal risk and quantitative acceptability criteria for evaluating levels of individual risk & societal risk. 4. Individual and societal risk metrics are assessing the impact of flood based on the number of human losses. The first is the probability of death for an individual in a certain location and the second relates the exceedance probability of an event and potential number of fatalities. The graphical representation of societal risk is an FN-curve, while of individual risk metric it is a number. The two metrics are complementary; a decision maker should have information about both metrics. 5. The first advantage of implementing these metrics in flood safety policy is that there is no need any more for monetization of human lives. Moreover, a decision maker has an overview of probabilities of events and their consequences, thus he may distribute resources and time available in a more effective manner. Finally, events with high rate of fatalities, which are undesirable and can cause disruption to the whole of the country, can be taken into consideration. Dimitris Detsis 4-11

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 6. Besides the estimation of individual and societal risk, setting and validating limits is advised. Legislation that defines certain risk for dike rings can be very effective. Of course, limits formulation cannot be based only on technical analysis but on society s needs and wishes. Finally, limits should calibrate effort required for protection measures with result on these measures. 4.7 References [4.1] Brinkhuis-Jak M. [et al.] Cost benefit analysis and flood mitigation in the Netherlands Heron. - 2004. - 1 : Vol. 49 [4.2] Eijgenraam C.J.J. Optimal safety standards for dike-ring areas - Discussion Paper. - The Hague : CPB Netherlands Bureau for Economic Policy Analysis, 2006. [4.3] Jongejan R. B., Jonkman S. N. and Maaskant B. The potential use of individual and societal risk criteria within the Dutch flood safety policy (part 1): basic principles. Reliability, Risk and Safety, ESREL / ed. R.Bris C.G. Soares, S.Martorell. - 2009. [4.4] Jongejan R.B. How safe is safe enough? The government s response to industrial and flood risks: Doctoral thesis / Delft University of Technology. - Delft, The Netherlands, 2008 [4.5] Maaskant B., Jonkman S. N. and Jongejan R. B. The use of individual and societal risk criteria within the Dutch flood safety policy (part 2): estimation of the individual and societal risk for the dike rings in the Netherlands. ESREL. - 2009. [4.6] van Dantzig D. Economic Decision Problems for Flood Prevention. Econometrica: The Econometric Society, July 1956. - 23 : Vol. 24. - pp. 276-287. [4.7] Vrijling J.K., van Gelder P.H.A.J.M. and Ouwerkerk S.J. Criteria for acceptable risk in the Netherlands / Infrastructure Risk Management Processes: Natural, Accidental, and Deliberate Hazards / book auth. Taylor Craig E. and Vanmarcke Erik.: ASCE, 2006. [4.8] Vrijling J.K., van Hengel W., l and Houben R.J. Acceptable risk as a basis for design Reliability Engineering and System Safety. Elsevier, January 1998. - 1 : Vol. 59. - pp. 141-150. 4-12 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report.5 Developments about seismic risk mitigation The objective of this chapter is to give a presentation of seismic risk mitigation process. Main principles of seismic risk estimation are going to be analyzed. The analysis will be focusing in the case of Greece and the developments there. Main sources of this chapter are the results of ongoing research programs in Greece, while references from E.U. or U.S. codes and guidelines are also studied. 5.1 Seismic risk Even in theory, the first step of risk management is risk identification. The common mistake is to associate risk with the hazard itself. Conversely, it is sometimes the case, that an area with increased seismicity may not be considered earthquake prone if loss after an earthquake event is low or the protection level is high. To define Seismic Risk it is necessary to look into its components: Seismic Risk is the result of the function of 3 elements (Figure 5-1) seismic hazard (likelihood of an earthquake and the associated severity), seismic vulnerability (expected damage to buildings given the occurrence of an earthquake), and expected consequences/losses. Expected losses can be casualties, capital, contents, business interruption, market share etc.(fema-389, 2004) Figure 5-1 diagram of seismic loss rough estimation for building stock in a region 5.2 Seismic hazard 5.2.1 Peak ground acceleration Earthquake is the ground shaking caused by the faults in the earth crust. Seismologists are measuring magnitude (M L ) of earthquakes using Richter scale, which quantifies the amount of seismic energy released magnitude. On the other hand intensity I (Mercali scale) measure qualitative the results of an earthquake on the Earth's surface, humans, objects of nature, and man-made structures on a scale of I through XII, with I denoting not felt, and XII total destruction. Dimitris Detsis 5-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Seismic hazard (SH) is the expected action on infrastructures and buildings caused by an earthquake; for engineering design SH can be described in terms of several parameters of motion like (peak or spectral) acceleration, velocity, displacement. One of them can be the value of the peak ground acceleration (PGA). It is not a measurement of the total size of the earthquake, rather than how hard the earth shakes in a given geographic area. Figure 5-2 gives an example of ground acceleration graph caused by an earthquake. Reference PGA (α g ), which is the peak ground acceleration (PGA) with 10% exceedance probability in 50 years in normal ground conditions, is used by building codes to prescribe seismic hazard for an area. The Figure 5-3 (Giardini, et al., 2003) Figure 5-4 (EPPO, 2003) show maps of seismic hazard for Europe as edited by European Geological commissions and for Greece as edited from Earthquake Planning and Protection Organisation. Figure 5-2 acceleration graph Figure 5-3 - European-Mediterranean Seismic Hazard Map (Giardini, et al., 2003) Figure 5-4 - Greek Seismic Hazard Map (EPPO, 2003) It can be observed that Italian & Balkan Peninsula, together with west coast of Turkey are the seismic prone areas amongst Europe and the Mediterranean Sea. 5-2 14 June 2010

5.2.2 Probabilistic Seismic Hazard Assessment Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report The above maps are results of probabilistic methods, are developed in order to estimate seismic hazard. These are described by the term probabilistic hazard analysis or assessment (PSHA). The goal of PSHA is to quantify the probability of exceeding various ground motion levels at a site (or a grid of sites) in a certain time interval, given all the possible earthquakes. Several ground motion parameters can be considered for PSHA as the maximum intensity at a site, the duration of the shock, the peak ground acceleration (PGA), the peak ground velocity (PGV), spectral accelerations (at structural periods of 0.2sec, 1.0sec, 2.0sec, etc.). The methodology of PSHA remains essentially the same in all cases. (Solomos, et al., 2008) PSHA steps are the followings: 1. specification of the models for the seismic sources responsible for the seismichazard; 2. specification of the ground motion models, i.e., the attenuation relationships; and 3. actual calculation of the sought exceedance probabilities Further analysis of this methodology is out of the scope of this study. However, for communication reasons, it is useful to mention some formulas and symbols, concerning the estimation of exceedance probability of α g. The probability of exceedance of a ground motion α g (reference peak ground acceleration) in the next T L (time span) years at the site will be: wtl P ( A > a ) = 1 e (5-1) or P A > a = e (5-2), where T g g L TL / TR T ( g g) 1 L w=w(a g ) annual rate of exceedance (1/yrs) T R =1/w mean or average return period The above equations are also known as the Poisson process. The following table shows the relationships between exceedance probabilities and corresponding return periods for a specific site representing scenarios of earthquakes. probability of exceedance time span T L (yrs) mean return period T R (yrs) 20% 10 years 45 years 20% 50 years 224 years 10% 50 years 475 years 5% 50 years 975 years 10% 100 years 949 years 5% 100 years 1950 years Table 5-1 examples of earthquake scenarios exceedance probability As mentioned, in building codes like EN 1998 and Greek Seismic Code (EAK 2000) it is recommended that the reference peak ground acceleration on type A ground ( good conditions), a gr, for the purpose of seismic zonation, corresponds to a reference probability of Dimitris Detsis 5-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering exceedance P NCR =0.10 in T L =50 years. This is equivalent to a reference return period of T NCR 475years for collapse limit state. For damage limitation, a time horizon of T L =10 years is considered, and the probability of exceedance is chosen equal to P R =P DLR =0.10. The above expression now produces a mean return period of T DLR 95 years. The following figure gives an example of PSHA for two different time spans. Figure 5-5 example of PSHA result (Solomos, et al., 2008) 5.2.3 Response spectrum To give better understanding on earthquake response, this paragraph outlines the parameters that influence seismic action on a building. In Eurocode 8 Part I (European Standard, 2003) and Greek Seismic Code -EAK2003 (EPPO, 2003) the earthquake motion at a given point on the surface is represented by an elastic ground acceleration response spectrum, henceforth called elastic response spectrum S e (T). The design seismic action on a building depends then on: peak ground acceleration soil conditions building s characteristics like: o vibration frequency o importance class o foundation s depth and rigidity o behaviour factor q The last parameter is introduced in order to avoid explicit inelastic structural analysis in design. Due to the capacity of the structure to dissipate energy, because of ductile behaviour the response spectrum is reduced. 5-4 14 June 2010

5.2.4 Collateral & indirect seismic hazards Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Besides ground motion, collateral damages caused by earthquakes must be mentioned here. These are surface fault rupture, soil liquefaction, or differential compaction, land sliding, and flooding. Building design codes are not intended to prevent damage due to damages caused by these hazards. For these cases, special risk management and cost benefit analysis is conducted. If this hazard is indentified in the site, and the relocation is not an option special design and engineering is necessary. This research project is not considering damages from those hazards. The assumption is reasonable since for the majority of the existing building stock in Greece the probability of occurrence is negligible. In addition to the above, there are also indirect hazards caused by an earthquake event. These are vulnerable lifeline systems, pounding and hazardous adjacent structures, storage, and distribution of hazardous materials and post-earthquake fires. 5.3 Seismic vulnerability 5.3.1 Definition The second component of seismic risk is vulnerability of buildings. Vulnerability, as defined by (TCG, 2001) is the potential of building to suffer damages. It expresses the expected damage given the seismic load. Figure 5-6 depicts a simplified vulnerability curve. Damage (D) is zero until a level of hazard (H) after that damage increases by the increase of hazard. Figure 5-6 simplified vulnerability curve The main factors that influence vulnerability of structures are (Karabinis, 2003): changes of reference PGA in the norms geology & geomorphology of the area design methodology and construction process maintenance Dimitris Detsis 5-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 5.3.2 Vulnerability estimation Current seismic design codes aimed at the prevention of life and safety for the benefit of the society. The approach used is limit state, so there is no view of structure s behaviour relation with earthquake intensity (and /or PGA). Consequently, even for a new designed building you can only recognize limit states as no damage or total damage. It is more obvious that vulnerability estimation is necessary for existing buildings because it can represent the behaviour of building considering the existing situation. Vulnerability is correlating seismic action/load with the damage degree of a building after an earthquake, while the later is defined as the percentage of seismic capacity reduction. Vulnerability can be relevant to the assessment of the structural behaviour of a building. Performance based design adapted by recently published intervention codes EC8 part III (Pinto, 2005) and KANEPE (EPPO, 2009) is the appropriate tool to associate damages caused by an action. However, detailed assessment of all buildings is a time consuming and costly procedure. For policy and decision-making in the urban or state scale, more qualitative approaches for the estimation are appropriate. The table below summarizes performance assessment methodologies. method description use empirical multi-criteria evaluation analytical classification of vulnerability depending on basic characteristics such age, building code, materials, structural system, ground etc scoring criteria with certain weights, those criteria are more detailed than the previous and focus on structural members, which are important for building behaviour (see appendix D) detailed structural analysis using structural modelling (see appendix D) Table 5-2 assessment methodologies (Karabinis, 2003) large sets of buildings seismic risk policy small sets of buildings preliminary check GO / NO GO for analytical assessment individual building mapping, analysis, retrofit design According to the above table, empirical methods are useful for seismic risk estimation of large building stocks of the state or of municipalities. Basic characteristics that are taken into account include the type and material of the structural system, design code, seismic hazard in the region, possible vulnerable construction practices like (ground soft floor, small columns and others), interaction with neighbour buildings, soil of the foundation. An empirical method of vulnerability assessment for seismic risk estimation of large sets of buildings is based on the following assumption: Expected damage is the same for all the buildings of the same category. 5-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Lately with the development of computers, there is the idea of creating a large database with the vulnerability of buildings, which will be measured with analytical methods. This is a reality for the U.S. (HAZUS) while efforts are being made in Greece for such a program. For the future creation of a database with the structural (and why not environmental) performance of buildings (either new or existing) will be an ultimate tool for macro and micro scale decisionmaking. Assessment of vulnerability of existing buildings is a critical step for seismic risk estimation. It actually encrypts the same difficulties as seismic risk mitigation itself. The implementation of a large-scale program for the assessment of vulnerability and then the seismic risk estimation requires government intervention, aiming either to collect data about macro-scale characteristics for all the stock or to ask owners to perform analytical assessment. Research programs in Greece (ARISTION & EPANTYK) have elaborated studies in order to model estimation of building vulnerability of Greek buildings (EPANTYK, 2007). On appendix E, this methodology is described. In general, this research program proposed 7 vulnerability curves for 7 distinctive structural typologies of buildings, relating damage degree with α g. In this MSc thesis, data and results of this study will be used, for a seismic risk estimation model. 5.4 Earthquake losses Losses of an earthquake event include human casualties and economic losses. The latter are distinguished to direct and indirect losses. 5.4.1 Human lives Seismic risk mitigation aims to protect human lives, beside economic losses. Thus, it is important for the risk estimation to include an assessment of the probable levels of human casualties, both deaths and injuries. However, it is a difficult endeavour because casualty numbers are highly variable from one earthquake to another and data-documenting occurrences of life loss in earthquakes is poor. Loss of life during an earthquake can be caused through many different ways: building total or partial collapse, machinery accidents, heart attacks etc. Also collateral and indirect effects (see par. 5.2.4 may increase also fatality number. (Coburn, et al., 2002) An approach to estimate these casualties is by determining the mortality ratio for each typology of building present in the stock damaged by an earthquake. Mortality ratio is defined as the ratio of the number of people killed over the number of occupants present in collapsed buildings of that class. Thus, to estimate fatalities it is necessary to know of collapsed buildings (or collapsed building area) of each class, and the mortality ratio for that class. Dimitris Detsis 5-7

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Mortality ratio is approximated using data from previous events. Factors that influence it are (Coburn, et al., 2002): building type and function occupancy levels type of collapse mechanism ground motion characteristics occupant behaviour and emergency operations effectiveness For the estimation of expected fatalities, several studies attempt to model correlating collapsed area with death / injuries or better damage degree of a building with a fatality rate (ATC-13), (Coburn, et al., 2002) & (Seligson, et al., 2006)). These kinds of studies should also depend on the typology of the building. Since this study did not find any mortality ratio (function) related with the building characteristics of Greece, the proposal for a research to be conducted on this field is repeated (also on the report of ANTYK, et al., 2001). Finally, similarly to floods, monetization of human life is still controversial. Kappos, et al., (2007) proposed the FEMA-227 (1992) courts award approach, where they found that indemnities reported varied from 50,000 to 1,450,000 for deaths after two different events in Greece. Later on, this study will propose risk metrics that do not require human life monetization. 5.4.2 Direct losses Direct losses include damages on buildings or infrastructure that can be caused by seismic action or collateral and indirect effects. Damages can be on the bearing structure but also in the architectural infill. Direct losses also regard contents of the building like fixed and movable equipment, goods for sale, etc. Direct losses are actually quantified as the costs required spending for the building to return to the previous state regarding seismic capacity and function. These costs include also parameters depending on the size and the importance (classification) of the building. Apart from the size of damage, the cost for repair is related to the structural typology, type of failure, repair materials and construction methodology, difficulty to access to the point of the failure, logistics in general and others. A correlation function between damage degree and rebuild cost can monetize direct losses from damages in buildings or infrastructure. For such a function to be formulated statistical data is used from repaired buildings. However, the estimation of losses from indirect hazards of an earthquake like fires, floods etc remain vague. 5-8 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 5.4.3 Indirect loses Indirect losses of an earthquake event can be sociological/ psychological, business interruption (loss of operation), loss in residential stock, unemployment, and others. Estimating these kinds of losses is very hard since it is difficult to predict or monetize them. It is true that knowledge about modelling estimation of indirect losses is missing. This does not mean that indirect losses are minor; on the contrary, they can be multiple of the direct (ANTYK, et al., 2001). Similarly, to what has already been discussed for the flood indirect losses, those of earthquakes are increasing non-linearly, depending on the intensity of the hazard (Figure 4-2). The same can be seen in the graph of Figure 5-7. The graph of Figure 5-7 (FEMA-389, 2004) shows losses related to their likelihood of occurrence. Potential losses (based on building value, building vulnerability and hazard level) are divided in direct and indirect (dotted line). For a low intensity-high probability event, direct losses are larger than indirect. However, when earthquake intensity increases indirect losses are multiplied. This finding is very important because, especially when seismic risk is dealt with in urban scale, the consideration of indirect losses might increase risks and give a different outcome to the decision-making. Figure 5-7 risk of experience catastrophic losses (FEMA-389, 2004) 5.4.4 Overview Figure 5-7 introduces the terms of manageable, catastrophic loss. Manageable loss refers to costs that can be covered from the state or in-house resources. If losses exceed resources, they are called catastrophic. Dimitris Detsis 5-9

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 5.5 Mitigation strategies Complete risk mitigation after risk determination and quantification would have proceeded to strategy identification step. The main categories are design, business, and event response strategies. The break down structure for these strategies is shown on Table 5-3: First cost design strategies reduce hazard relocate to better soil relocate to lower seismic zone engineer soil material or seismic isolation reduce vulnerability reduce response mass: passive/active dampers increase seismic capacity more strong or more ductile Operating Cost diversify operations geographical spread Business Strategies insurance obtain higher insurance level securitization catastrophe bond (cat bond) instruments Event response emergency response procedures and materials strategies pre-event disaster training and inspection emergency operation centres Table 5-3 seismic risk mitigation strategies These strategies refer to owners of buildings, but can be extended to a stock of buildings and used by policy makers. First cost or design strategies refer to methods to reduce likelihood of damage (reducing hazard or vulnerability). Operating cost or business strategies aim to reduce the loss, these can be spreading operations (available only for owner with multiple properties) and/or insure seismic risk. Securitization refers to the use of financial instruments to transfer risks to capital markets. While an estimated loss of i.e. 35 billion euro would be an unparalleled loss for the traditional (re)insurance industry, it would be business as usual for the global securities market (stocks and bonds) where daily fluctuations normally exceed this figure by far (Jongejan, 2008). Finally, event response strategies are triggered at the time and after the occurrence of the event and aim to reduce losses. Reduce vulnerability (retrofit) strategies, especially by increasing capacity of structure, has a range of possible solutions and costs. Hence, it is often employed because it allows the engineer to fine-tune a design approach to meet both an owner s budget and the risk management criteria. However, it is difficult to quantify their total cost (FEMA-389, 2004). Furthermore, the chosen strategy might be the optimal combination of the above. 5-10 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 5.6 Retrofit strategies 5.6.1 Designing retrofit Design decisions for the engineer for an individual existing building are: Repair Strengthen / retrofit Demolish/reconstruct The decision regarding demolishing/reconstructing depends on the costs of intervention, but it is also a function of other factors, like building terms and historical/archaeological value. In the following table, one can see different types for retrofit existing building as recommended by Earthquake Planning and Protection Organization (EPPO, 2001): How? Where? examples TYPE I increase ductility existing elements thin mantle in columns capacity & connections wrapping with sheets / frp TYPE II increase strength / existing elements increase shear walls thickness stiffness & connections TYPE III increase strength / existing elements increase shear walls thickness stiffness and ductility & connections and mantles in columns TYPE IV increase strength / new elements add new shear walls thickness stiffness and ductility and mantles in columns TYPE V embedment of passive damping system Table 5-4 Retrofit strategies for r/c buildings (in Greece) The influence of these types on the behaviour of the building can be visualized in V-δ, where V is the horizontal load on the base of the building (proportional to hazard level) and δ is the displacement of the roof (proportional to building response). The bold parabolic line depicts seismic demand and the other curves depict behavior of the building. If these curves are above seismic demand then the building is safe. The dotted line shows the reduced seismic load after seismic isolation of the building (TYPE V). Figure 5-8 characteristic capacity curves for retrofit strategies (EPPO, 2001) Appendix G gives more figures relevant with Table 5-4. Dimitris Detsis 5-11

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 5.6.2 Retrofit costs and benefits For an individual building, costs and benefits for retrofit can be summarized in the following: Costs direct indirect investigation and assessment of existing situation design / permits for retrofit project construction cost loss during construction and / or cost for permanent accommodation business interruption Benefits (from less risk) direct less direct losses after an event indirect less indirect losses after an event increase of value and quality of the building Table 5-5 costs & Benefits of building retrofit For the owner there are several reasons not to want to invest in retrofitting his building, such as short-time horizon, desire for a quick return of investment, has budget constraint or lack of perception of added economic value.(grossi, 2008) 5-12 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 5.7 Conclusions 1. Main components of the seismic risk estimation for the building stock are hazard, vulnerability, economic and social characteristics of buildings. In Greece, seismic hazard is prescribed by the latest Seismic Hazard Map, which is a result of probabilistic seismic hazard assessment. Vulnerability relates damage with hazard. 2. Vulnerability of buildings can be measured using empirical multi-criteria, and analytical methods. Latest development is the estimation of vulnerability of buildings by various research programs in Greece using empirical methods and statistical results (data from previous events) categorize them into different structural typologies. 3. Expected losses after an earthquake are human casualties, direct (economic) loss (damage in buildings and infrastructure, loss of contents), and indirect loss. Regarding the first, there are qualitative estimation methodologies, not specialized though for the case of Greece. For direct losses, several studies are relating damage degree with cost of replacement; however, it is necessary to know the size & the importance of a building. Regarding the later, although inventory data have been gathered for all buildings of Greece, the important parameter of the building area is missing. 4. Indirect losses are very difficult to estimate but very important. Indirect losses, like sociological/ psychological, business interruption (loss of operation), loss in residential stock, unemployment etc increase exponentially as the hazard increases. 5. The main categories of mitigation are design, business, and event response strategies. Design or first cost strategies are those that reduce hazard and/or vulnerability, therefore reduce probability of damage. Operating cost / business strategies diversify or insurance instruments and event response strategies aim to reduce consequences. 6. Greek authorities have already published a new building code and documents to support technically the design and implementation of retrofit projects. Design strategies of retrofit are explained; those are either combination of increasing strength, stiffness, and ductility on existing elements or adding new elements or seismic isolation of the structure. Cost for design & construction of these, along with loss of rent and building interruption during construction time have to counterweight the benefits from the reduction of risk. Regarding the benefits, the increased value and quality of retrofitted building should not be forgotten. Dimitris Detsis 5-13

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 5.8 References [5.1] ANTYK, Theodorakis S. and Vougioukas E. Priorities policy for seismic retrofit / Seismic Retrofit of Existing Building (ANTYK) - Athens:Technical Chamber of Greece, 2001. GREEK. [5.2] Coburn A. and Spence R. Earthquake Protection - 2nd. : John Wiley & Sons Ltd. [5.3] EPANTYK Pre-seismic Retrofit of existing buildings [Book] / ed. Buildings National Program for Seismic Retrofit of Existing: Technical Chamber of Greece (T.C.G.), 2007. - GREEK. [5.4] EPPO Greek Seismic Code / ed. (OASP) Earthquake Planning and Protection Organization. - Athens : Ministry of Environment, Planning and Public Works., 2003. - BUILDING CODE - GREEK. [5.5] EPPO Code of Interventions (KANEPE)/ ed. OASP Earthquake Planning and Protection Organization: Ministry of Environment, Planning and Public Works, 2009. - GREEK - BUILDING CODE. [5.6] EPPO Recommendations for pre-earthquake and after-earthquake intervention in buildings /ed. OASP Earthquake Planning and Protection Organization, Athens : Ministry of Environment, Planning and Public Works, 2001. - GREEK. [5.7] European Standard Eurocode 8: Design of structures for earthquake resistance - Part1: General rules, seismic actions and rules for building - 2003. - pren 1998-1. [5.8] FEMA-389 Primer for Design Professionals Communicating with Owners and Managers of New Buildings on Earthquake Risk : Federal Emergency Management Agency, 2004. [5.9] Giardini D., Jimenez M.J. and Grunthal G. European-Mediterranean Seismic Hazard Map: European Seismological Commission, 2003. [5.10] Grossi P. Modeling Seismic Mitigation Strategies / Risk Assessment, Modeling and Decision Support / book auth. Bostrom A, French S and Gottlieb S.: Springer Berlin Heidelberg, 2008. - Vol. 14. - 10.1007/978-3-540-71158-2. [5.11] Jongejan R.B. How safe is safe enough? The government s response to industrial and flood risks: Doctoral thesis / Delft University of Technology. - Delft, The Netherlands, 2008 [5.12] Kappos A. J. and Dimitrakopoulos E. G. Feasibility of pre-earthquake strengthening of buildings based on cost-benefit and life-cycle cost analysis, with the aid of fragility curves / Natural Hazards. - August 21, 2007. - Vol. 45. - pp. 33-54. [5.13] Karabinis A.I. Assessment of seismic behaviour of r/c constructions Vulnerability and Risk / 14th Concrete Conference. - Kos, Greec :2003. [5.14] Pinto E.P. The Eurocode 8 Part 3: The New European Code For The Seismic Assessment Of Existing Structures / Asian Journal Of Civil Engineering (Building And Housing). - 2005. - 5 : Vol. 6. - pp. 447-456. [5.15] Seligson H. A., Shoaf K. I., Kano M. Development Of Casualty Models For Non-Ductile Concrete Frame Structures For Use In Peer s Performance-Based Earthquake Engineering Framework /8th U.S. National Conference on Earthquake Engineering. - San Francisco, California, USA, 2006. - Vol. No. 917. [5.16] Solomos G., Pinto A. and Dimova S. A Review Of The Seismic Hazard Zonation In National Building Codes In The Context Of EUROCODE 8: Support to the implementation, harmonization and further development of the Eurocodes. - Brussels : JRC - European Commission, 2008. - EUR 23563 EN - 2008. [5.17] TCG - Staff Commitee Seismic retrofit existing buildings. Summary of results of 1st phase of research project in G.T.C. - Athens : Technical Chamber of Greece, 2001. - GREEK. 5-14 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report.6 Conclusion of research phase This chapter concludes the research phase. After looking into main elements of hazard risk management, criteria for the assessment of existing buildings, cornerstones of Dutch flood risk management practices and the latest development regarding seismic risk mitigation in Greece, the question which elements of flood risk management are suitable for seismic risk mitigation is answered. 6.1 Stimulating seismic risk mitigation 6.1.1 Government Intervention As discussed in paragraph 0a way to stimulate seismic risk is government intervention. After problem analysis of chapter.1, and close examination of recent development regarding seismic risk mitigation (in chapter.5 and appendix H) it came out that experts in Greece also ask for government intervention for the creation of seismic safety plan. However, central or local governments are not yet responding to undertake pre-seismic mitigation planning (see problem analysis of chapter.1) A government intervention, similar to strategies referring to cases of environmental hazards discussed in paragraph 0could consist of: a safety plan implemented by government, which may includes: i. retrofit program, ii. mandatory insurance, iii. emergency planning etc incentives like subsidies, tax reliefs etc introduction of more stringent building codes risk communication 6.1.2 Centralized concept Government intervention may be organized in two levels, top level (central administration/government), and medium level (local administration/governments). According to the above strategies and the main risk management decision steps, the following box and Figure 6-1 describes the environment of this intervention. Dimitris Detsis 6-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Top level - Central administration 1. estimates seismic risk for regions (municipalities) and for total country, 2. outlines safety plan (re-estimation dynamic model) 3. intervention (to local administration and to owners) 4. monitoring Central administration outlines safety plan deciding upon general risk management actions for each municipality: accept insure mitigate (retrofit, de-urbanize, improve emergency plans) Medium level - Local administrations 1. estimate seismic risk for every building typology and total area, 2. draw local safety plan according to top level intervention 3. intervention (to owners) 4. monitors and gives feedback to central administration Local administrations draw safety plan according to central administration intervention. Their options can be: organize retrofit programs draw local emergency plans Low level Building owners 1. estimate risk (building seismic performance assessment) 2. decide mitigation strategy (see Table 5-3) 3. execute 4. monitor Buildings owners will have to choose from mitigation strategies of Table 5-3 taking into account (in order of priority) intervention of central or local administration cost / benefit analysis Table 6-1 risk management steps for a government intervention 6.1.3 Need for risk metrics What is obvious on the above scheme is that risk estimation plays a significant role on the execution of government intervention. Whatever alternations the above scheme takes and for every scheme of government program, risk estimation is vital to be able to set priorities and decide whether buildings, municipalities, or regions are safe enough. However, risk estimation instruments, which can help for such a centralized decision-making, have not been developed yet. Comparable to tools described in chapter.3, which refer to individual buildings, and to risk evaluation efforts in Greece referring only to the earthquake design scenario (EPANTYK, 2007), the risk for centralized decision-making should have a more societal manner and be able to help setting priorities and then risk acceptance criteria. Moreover, besides economic risks, risk to life should also be considered separately. 6-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Figure 6-1 - decision making scheme A review of the cornerstones of Dutch flood risk management practices in chapter.4, especially in risk estimation and decision-making, has shown that fatality risks are considered from a societal perspective and an individual one. Two different metrics are developed: the societal risk metric concerns the (exceedance) probabilities of larger numbers of fatalities; the individual risk metric concerns the probability of death of a person at a specific location. The concept has been extended from Dutch Major Hazard Policy. Chapter.5 shows that elements are already available. Thus, the calculation of the above metrics for the case of seismic risk in Greece may be possible. Finally, the analysis of next paragraph, about forced centralized decision making in The Netherlands regarding floods, strengthens the idea of using theses metrics for the case of seismic risk. Dimitris Detsis 6-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 6.2 Position of defence line in flood protection In the case of flood risk, dikes & levees are surrounding polders in which urban development takes place and they form a defence line. Central administration is responsible for safety plan, so it designs and constructs these dikes and levees. All the inhabitants decide their level of safety. This gives the advantage of collective knowledge and experience, which can be prove more effective in mitigating the hazard. Of course, there are complexities in the decision-making procedure since every individual has different perception of risk, but in the end, every individual inside a polder is equally vulnerable to flood. Moreover, as discussed in chapter.4, there is an effort to deaden oppositions of individuals or society, regarding the risk levels by the application of risk acceptance criteria. The figure is different in the seismic risk case. Technically, in this case, defence line is in every unit (building / asset) of the urban block (city, prefecture etc). Therefore, protection against hazard is in the hands of building owners. It is a non-centralized process for hazard protection/ mitigation. The level of seismic protection of assets in an urban area is not equally distributed, since the urban block contains buildings of different vulnerability due to age, poor construction, or maintenance etc or even because owners are of different socio-economic situation. For example, it is notable that poor districts proved to experience larger consequences. Finally, even if the owner has the ability to pay, he/she may have a different perception of risk; a risk-averse owner will be willing to increase safety of his/her property while a risk seeker will accept more risk. The result is the creation of different levels of safety inside the same urban block (municipality or prefecture) and within the whole country. This is translated to different life expectancy and different wealth safety. The different concepts of those two cases are depicted in the following Figure 6-2. Technically the concept of outer defence line of flood risk mitigation cannot be extended - it is not possible to seismic isolate an urban block. However, government intervention would end up to the centralization of the process and could stimulate seismic risk mitigation. This could improve the distribution of resources and knowledge, help for smooth decision making upon priorities or criteria. The idea is to shift from individual decision making to societal estimation and mitigation of seismic risk in order to reduce impacts of earthquakes, in general, for a region or for the whole country. 6-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Figure 6-2 different concept of the two cases Dimitris Detsis 6-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 6.3 Conclusion In the case of the Netherlands, due to the nature of the flood hazard and protection scheme (public flood defences), the government is strongly involved in flood risk mitigation. Despite the differences between the protection schemes for large-scale floods (strengthening dikes rather than protecting buildings) and earthquakes (strengthening buildings), this project proposes the translation (=the act of converting) of societal risk and individual risk as instruments of seismic risk estimation in Greece. This translation seems, according to the previous chapters, to have promising potential firstly because it discriminates fatalities from other economic losses. Secondly, it is an integrative methodology (similar to the concept described in Figure 2-1 in chapter.2), which helps organize a safety plan and have an overview of its results. Finally, as described in chapter.3, regarding assessment criteria of (existing) buildings, there is a philosophy change to performance-based design for structures when it comes to natural hazard protection. The current proposal extends this integrative philosophy to the upper level of decision-making, from the structural safety of a building to a safety plan for a region or country. Part III of this report deals with modelling (in the frames of this translation) of societal risk and individual risk metrics for the case of Greek seismic risk mitigation. 6.4 References Contents of this chapter depend on previous chapter analysis! 6-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report PART III SEISMIC RISK ESTIMATION M O D E L I N G.7 Description of developed estimation model The third part of this research project is based on the application of use of societal & individual risk for seismic risk safety policy in Greece. Besides that, the application worked out combines the calculation of direct economic loss for different earthquake scenarios. Societal risk (FN-curve), individual risk (AEF), direct economic risk (FL-curve) and annual expected economic loss (AEL) can be handy instruments for decision-making. Chapter 7 outlines general philosophy of this project s proposal, presents the methodology, discusses the input data used, and runs through indicative results. Then it presents the way mitigations are modelled. 7.1 Philosophy 7.1.1 General description The proposal of this research is a model that mainly estimates societal & individual risk for existing building stock of municipalities of Greece. Then with a similar method, direct economic loss is estimated for various seismic scenarios. Moreover, mitigation strategies (mainly building retrofit) are modelled in order to quantify the potential of these strategies and compare their effectiveness. The methodology described in this chapter derives from the four risk metrics estimators, outlined in Table 7-1, for a municipality/region in Greece for the existing or retrofitted stock of buildings. Within the frames of this project, besides retrofit of building stock, the influence of deurbanization on seismic risk reduction can be evaluated for cities of Greece. Finally, a rough cost-benefit analysis model finds the maximum retrofit cost for a feasible investment. Additionally intermediate outputs can be also useful. 7.1.2 Restrictions The seismic risk estimated by the developed model refers only to fatalities or economic loss caused by damage in buildings from the seismic action. Fatalities or economic losses due to the following are not taken into account: collateral events of the earthquake in general damages in buildings due to collateral events damages to infrastructures Dimitris Detsis 7-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering human lives risk metric relates exceedance probability of an event and potential number of fatalities 1) societal risk 2) individual risk 3) societal economic loss* PN area area ( all nall ) OR FN ( ) probability of death of an average, unprotected person that is constantly present at given location (location is either building or municipality) relates exceedance probability of an event and potential economic loss PL all area area ( ) l all OR FL () outcome 3) FL-curve*** 2) AEF/per capita** 1) FN-curve economic wealth 4) individual econ loss* annual expected economic loss per citizen in location (location is either building or municipality) 4) AEL/per capita** * Only direct economic loss is considered at this study ** Maps are presented here, only indicative here, not in the final result of this project *** Different shape of curves of 1 & 3 is due to different deterministic functions based on which they are calculated note F: annual exceedance frequency (equivalent with annual exceedance probability P R,1) Table 7-1 main output of the proposal As far as economic losses are concerned, (FL-curve) only direct loss is taken into account here. However, the philosophy and the methodology would not change, if other indirect economic losses of an earthquake or casualties from collateral events or damages to infrastructure were to be included. As already mentioned, the aim of this project is to outline the philosophy and methodology of an estimation process for the proposed metrics, then explain their position and utility in the decision making process, evaluate their potential and give recommendations for further research or development. To do that, results for seismic risk of regions in Greece are presented here (please see Disclaimer clause). 7-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 7.2 Modelling risk estimation Societal risk is graphically represented by an FN curve which shows annual exceedance probability (or exceedance frequency) of the potential number of fatalities. Respectively, direct economic loss is depicted by an FL-curve showing annual exceedance probability (or exceedance frequency) of the potential direct economic loss. 7.2.1 Theory Theoretically, (and not exactly in this study) to quantify the earthquake risk for a building of a certain typology in a certain region, the following must be combined: 1. probability density function of an earthquake event in the region (probabilistic hazard assessment PSHA), 2. probability density function of damage degree on a building of this area given the earthquake event (fragility curves), 3. expected fatalities or direct economic loss given the damage in the building (mortality and loss functions) These are enough to calculate fatalities and direct economic loss per m 2 of one building of certain typology. To find the result for a region, inventory data is necessary. The necessary inputs should include the surface area of buildings of each typology, their importance (consequence class) and their inhabitants (occupancy rate). The following box presents the mathematical background: The exceedance probability of loss is calculated using the following: ( l ) P >L 1 p( ) L = l dl(7-1) 0 Equation (7-1) can be solved by defining discrete number of scenarios (i=1,2,..,n) with probabilities Pi and associated li. Scenarios should be mutually exclusive and collectively exhaustive. Finally, expected losses can be calculated as following: Ε = l p() l dl(7-2) () l 0 which also can be solved numerically for discrete number of scenarios () l n Pi l (7-3) i i= 1 Ε = Dimitris Detsis 7-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 7.2.2 Practical application for this study The model created (in excel) for the purposes of this study is based on the following: exceedance probability function of α g for a location (PSHA researches see par.7.3.1 ) is transformed to exceedance probability function of loss (societal risk) P (A a ) P(L l) (7-4) g g using deterministic transfer functions to calculate loss (fatalities or economic loss) given the value of α g with a certain exceedance probability (see par.&) g f d ( a g ) f l ( d) α d l (7-5) calculation is numeric in excel, using 14 different values (j=1,2,..,14) of annual exceedance probability (or annual exceedance frequency) annual expected loss is approximated using the following formula: () l Ε = P l n j= 1 where P = P( l > L ΔL) P( l > L +ΔL) j j (7-6) j j j The following table summarizes the main inputs for the calculation of an earthquake event consequences associated with their exceedance probability. input description symbol units annual exceedance probability value F 1/year Hazard PGA with certain annual frequency ag m/s2 Vulnerability ag damage degree of certain typology given the d = f ( a ) % i d g Mortality fatalities of certain typology given the damage N ( ) i = fn d ι people Loss function econ. loss of certain typology given the damage L = f ( d ) ι l ι Inventory data population, number of buildings per typology etc Buildings average area Occupancy rate A O buil. area Table 7-2 main inputs 7-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 7.2.3 Diagram The following diagram depicts the calculation process: Figure 7-1 diagram of the calculation methodology Dimitris Detsis 7-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 7.3 Main elements of calculations 7.3.1 Seismic hazard Exceedance probability function of PGA or α g for a location is estimated using the formula of Papazachos, which was found in the report of Papaioanou, et al. (2008): log10y=c1 log10t R +C (7-7) 2 Y Τ R c 1,i c 2,i :motion parameter here PGA :return period :different coefficient for 3 zones of seismicity considered in national seismic code (EAK,2003) i =seismic zone I,II,III The formula is used for the zonation of Greek territory. The following graph depicts annual exceedance probability of certain value of reference PGA (or α g ) for 3 different seismic zones. While Table 7-3 values of PGA or α g with their annual exceedance frequencies are gathered. P( A > a ) = P ( a ) = F( a ) (7-8) g g R,1 g g Figure 7-2 hazard curves for 3 seismic zones of EAK2003 It is reasonable to assume, when the stocks of buildings are examined, that for the area of a municipality (less than 40 s.km.) PGA will be the same for all the buildings. This assumption is in line with the research about PGA estimation after an earthquake. Even then, values are given per municipality (Elenas, 2003). 7-6 14 June 2010

annual time return peak groun acc. (a exceedance g ) 1 span period frequency log 10 Y=C 1 *log 10 T R +C 2 P R T L T R P R,1 or F ZONE-I ZONE-II ZONE-III 1% 10000 994992 1,0E-06 0,93 1,24 1,66 1% 1000 99499 1,0E-05 0,55 0,76 1,05 1% 100 9950 1,0E-04 0,32 0,46 0,66 2% 100 4950 2,0E-04 0,28 0,40 0,57 2% 50 2475 4,0E-04 0,23 0,34 0,50 10% 100 949 1,1E-03 0,19 0,28 0,41 10% 50 475 2,1E-03 0,16 0,24 0,36 20% 50 224 4,5E-03 0,14 0,20 0,31 10% 20 190 5,3E-03 0,13 0,20 0,30 20% 20 90 1,1E-02 0,11 0,17 0,26 50% 50 72 1,4E-02 0,10 0,16 0,25 2% 1 49 2,0E-02 0,10 0,15 0,23 50% 20 29 3,4E-02 0,08 0,13 0,21 10% 1 9 1,0E-01 0,07 0,10 0,16 50% 1 1 5,0E-01 0,04 0,07 0,11 Table 7-3 annual exceedance frequencies of PGA (*g=10m/s 2 ) 7.3.2 Vulnerability of buildings Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report As mentioned before, vulnerability is the potential of buildings to suffer damage under the seismic action. For the current study, vulnerability data for buildings in the territory of Greece is acquired by the result of a research program elaborated by TCG (EPANTYK, 2007). According to this research, buildings in the Greek area are classified initially according to: main construction material & morphology of bearing structure, design code at the time of construction number of floors (height), place and soil conditions However, it came out that for seismic risk estimation of large stocks of buildings, it is possible to distinguish 7 structural typologies taking into consideration the material, design code and soft ground floor existence. The following table shows these structural typologies: code (for this study) material design code pilotis (ground soft floor) 1 MS masonry all yes & no 2 RCOLN reinforced before no 3 RCOLY concrete 1985 yes 4 RC85N 1985 no 5 RC85Y 1995 yes 6 RC95N 1995 no 7 RC95Y today yes Table 7-4 main structural typologies according to (EPANTYK, 2007) Dimitris Detsis 7-7

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Further analysis and documentation about vulnerability as well as a short description of the structural characteristics of these typologies can be found in the appendices E & F. The pre mentioned research program provided vulnerability curves for these 7 typologies, relating their damage degree with α g /α o the fraction of PGA (α g ) that acts on the structure, to α 0 the reference PGA provided by today s seismic code). Damage degree is defined as the fraction of seismic capacity of the building after damage to the seismic capacity before. For this research, the curves are fitted (using regression analysis) to a quadratic function. Thus for every typology ( d = f ( a ) i g 2 ag ag i = 1, i + 2, i + 3, i a0 a0 d c c c i ), a transfer function of the following form is considered: (7-9) i d a i g a c 0 1,i c 2,i c 3,i : building typology code (i=1,2,,7) : damage degree for building typology i (%) : PGA for type A - normal soil conditions (g) : building s code design reference PGA (g) coefficients derived by regression analysis to vulnerability curves acquired from (EPANTYK - TCG, 2007) The result for each typology is shown in the figure below: Figure 7-3 vulnerability curves for each typology (fitted to 2 nd polynomial) Damage for every seismic scenario is deterministically calculated. Figure 7-4 is depicting exceedance probability curves for each typology for seismic zone I (α 0 =0.16). 7-8 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Thus, instead of using fragility curves, which are usually used in similar studies, this project estimates exceedance probability of damage degree by relating exceedance probability of a g (hazard) and the damage caused by this hazard. Figure 7-4 annual exceedance probability of damage for a building of each typology 7.3.3 Mortality function Next main element for the calculation of fatalities is the function relating the damage in the building to the fatalities. Due to the absence of probability density functions of fatalities or direct economic loss given the damage, mortality function is also considered deterministic. Coburn & Spence (2002) propose the following formula relating number of fatalities with the collapse area. K s=a col [M1 M2 M 3 (M 4+M 5 (1-M 4))] (7-10) (Coburn, et al., 2002) K s Α col M 1 M 2 M 3 : casualties : area collapsed : occupancy rate of building area : coefficient of occupants in building : coefficient of occupants trapped M : coefficient of deaths at t=0 4 M 5 : coefficient of deaths at t=36h Excluding M 1 at this point of calculations (later it is going to be considered) mortality function is formulated with the meaning of ratio of fatalities per m 2 of built area. After estimating the percentage of collapsed area (theoretical value) for each damage state using engineering reasoning and giving values to coefficients (M 2, M 3, M 4 & M 5 ) according to the data provided by Coburn & Spence (2002), the mortality function is formulated, see Table 7-5. N ( ) i = fn d ι (7-11) Dimitris Detsis 7-9

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Coefficient M 2 (ratio of people inside the moment of the earthquake) depends on the character of the area (rural or urban), and the governing function of the buildings. The M 3, M 4 and M 5 coefficients depend on the characteristics (height, material) of the buildings, for example for Athens with 100% urban character, 83% residential function, average height of 3,3 floors and 90% r/c buildings the values for these coefficient are: M 2 =0,57 M 3 =0,65 M 4 =0,40 M 5 =0,70 These numbers are considered the same for all the buildings independently of their typology. Detailed determination of this coefficient does not influence the result in a large scale (see paragraph 8.1.4 ) Finally, Table 7-5 and graph of Figure 7-5 actually give the mortality function, if the above coefficients are used. DAMAGE DEGREE F collapse CASUALTIES state description range mean area* S.Injuries Fatalities 0 DS-1 slight 0% 1% 1% 0,001% 0,0004% 0,0003% 0,0003% DS-2 moderate 1% 10% 6% 0,01% 0,004% 0,003% 0,003% DS-3 heavy 10% 30% 20% 0,1% 0,04% 0,03% 0,03% DS-4 v. heavy 30% 60% 45% 1% 0,4% 0,3% 0,3% DS-5 collapse 60% 100% 80% 10% 4% 3% 3% Table 7-5 mortality in different damage stage Figure 7-5 mortality in different damage states To increase validity of mortality function two more references are checked. The scale of magnitude for the mortality rate is similar to numbers given in ATC-13 (1985) and other works like (Seligson, et al., 2006) and (ANTYK, et al., 2001). However, since neither of the used studies is specifically referred to the case of Greece, it would be useful if such a research was conducted in the future. 7-10 14 June 2010

7.3.4 Direct economic loss function Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Direct economic loss has two components: repair costs to restore the previous situation of the building and loss of contents. First, loss of contents is considered proportional to 1/3 of the damage multiplied by the reference rebuild cost RC (Kappos, et al., 2007) (see equation 7.12). The other component repair costs, is related to damage degree through a 3 rd polynomial multiplied by the rebuild cost. This relation is acquired through a research, regarding data from repaired buildings after earthquakes (Kappos, et al., 2006). Since this data contains only cases of buildings that were to be repaired, the function is calibrated to give lower values for minor damages (Figure 7-6). Li = f ( di) 3 2 1,65-1,70 +1,20 +0,01 RC 1 (7-12) Li = di di d i i d RC + 3 i i BUILDING _ REPAIR _ COMPONENT CONTENTS _ COMPONENT i Figure 7-6 repair cost function For a damage of 0-1%, costs are forced to zero, while for a damage degree of 1% - 5%, although no repairs are decided, some assessment cost (engineer fee and/or investigation works) should be considered. For an 1% damage in which only the engineer fee is required, an average of 22 /m 2 is a reasonable price. See also paragraph 8.1.6 The value for rebuilding cost RC - is 800/m 2. This is an indicative value of dwellings construction cost in Greece. i Dimitris Detsis 7-11

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 7.4 Inventory data 7.4.1 Social numbers & existing building stock Greek National Statistical Authority (GNSA) provided inventory data for regions (municipalities or prefectures) of Greece. Data was collected on 2000-2001 and it is available on the website of GNSA. The following are considered necessary for the seismic risk estimation in a municipality: population, municipality area, number of buildings, height of buildings (floors) to estimate average building area, material, date of construction, pilotis to determine typology, character (urban or rural) to determine mortality rates, buildings function to determine mortality rates 7.4.2 Average buildings area Important data for the calculation of direct economic loss is the buildings average area. Unfortunately, this data is not available. To reach a satisfactory level of estimation for this parameter, for the needs of decision making of this scale, the following procedure was followed: Estimation of footprint of buildings by multiplying municipality total area with urban coverage factor (for Athens - 50%) calculation of average nr of floors (data available) calculation of average building area (by multiplying the above two) The following symbols are noted here: A floor A buil. O area : average floor area in the examined region : average building area in the examined region : occupancy rate of the examined area (m 2 /people) It is true that masonry buildings have a smaller area than the r/c buildings. Because of absence of specific data but also due to the minor influence on the result, regarding large stocks, like those of major municipalities, the average building area is considered equal for all typologies. In addition, missing data were those referring to the classification of the buildings (importance class). However, since stocks examined contain dwellings with a ratio over 80%, neither this parameter is going to be taken into consideration. It is notable that the number of fatalities, thus societal risk and individual risk, does not depend on the average buildings area. Only the number of buildings and the population influence the fatalities number. 7-12 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 7.5 Example The following example on calculating the number of fatalities and the direct economic loss with certain exceedance probability makes the calculation process more clear. For annual exceedance probability P R =10% in T R =50yrs F=0,0021 1/yr Calculation of the number of fatalities with exceedance probability F=0,0021 1/yr: 1. seismic zone I α 0 =0,16 a g =0,16 (equation 7-1) 2. α g /α 0 = 1,0 d 1 =19% (equation 7-3 for i=1-ms-masonry) 3. d 1 =19% heavy damage state M 1 =0,03% (table 7-4) sm sm 6 4. N1 =Μ 1 O area N 1 = 4 10 fatalities/m 2 5. buil. sm buil. 3 N1 = N1 A buil. N1 = 4 10 fatalities/building. 6. area buil area N1 = N1 nr 1 N fatalities 1 = 39 Repeat steps 1-6 for every typology (i=2,3...7) 7. 7 area area N = N = 56 _ fatalities all i= 1 i Calculation of direct economic loss with exceedance probability F=0,0021 1/yr: 1. seismic zone I α 0 =0,16 a g =0,16 (equation 7-1) 2. α g /α 0 =1,0 d 1 =19% (equation 7-3 for i=1-ms-masonry) 3. d 1 =19% L 1 =200 /m 2 (equation 7-6) 4. buil. sm buil. L1 = L1 A buil /building. L 1 = 22881 area buil. area 5. L1 = L1 nr 1 L1 = 2340 _ million _ Repeat steps 1-5 for every typology (i=2,3...7) 6. 7 area area L = L = 5682 _ million _ all i= 1 i Repetition of the above procedure for different values of exceedance probability Drawing FN-curve PN ( area all area area area n ) and FL curves PL ( l ) all all all The example continues in calculating individual risk metrics (Example: Athens) Annul expected fatalities per capita in an area (average individual risk): AEF AEF / capita = = = 5,5 10 population population n area Pj ( Nall ) j j = 1 6 Maximum annual expected fatalities in a building of an area (max individual risk): MAX n buil. Pj ( Ni ) j j = 1 5 inhabitants i, = 2,4 10 Dimitris Detsis 7-13

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 7.6 Output / Results 7.6.1 Societal risk (FN curve) & Societal economic risk (FL curve) Figure 7-7 - societal Risk for all area & for each typology Figure 7-8 societal economic risk for all area & for each typology 7-14 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report The societal risk gives an overview of exceedance probability of certain number of fatalities. The representation is the FN-curve. The x- axis stands for the number of fatalities N and the y-axis for the annual exceedance probability / frequency F of n number of fatalities, [1/yrs]; both axes are logarithmic (Figure 7-7). The range of the graph is from annual exceedance frequency F=0.20 (return period T R =4yrs) to F=10-5 (return period T R = 100000yrs). Examining such frequent events gives the opportunity to assure that the least safety is ensured for the population living in the existing stock for a short time horizon. Furthermore, it indicates if immediate measures should be considered. However, estimation in such frequent events is hard to be accurate. On the other hand, a return period of 100000 years seems to be unreasonable to consider, but estimation of losses in such low-frequency events is useful for insurance policy. Besides, as already mentioned, the main role of societal risk is to a give an integrative risk view of probabilities and the results of the hazard. The graph also shows societal risk for the stock of buildings of the same typology. Similar with societal risk, societal economic risk is presented in FL-curve, where F stands for frequency and L for losses (Figure 7-8). 7.6.2 Individual risk metrics Individual risk, as described in chapter.4 and estimated for floods in The Netherlands, is defined as the probability of death of an average, unprotected person that is constantly present at a given location. In this situation, two numbers are calculated. One represents the annual expected fatalities for a municipality, named average individual risk, which is an average number indicating the risk for a person living somewhere in the municipality. Calculating these numbers for all the municipalities, a risk map can be completed. The second stands for the maximum annual expected fatalities, named maximum individual risk, per municipality. This provides an overview of disproportional exposure of population of municipalities. However, since used data roughly include the assumption that all the buildings inside a municipality have the same size, maximum individual risk is only given suggestively. More analytical methods applied on each building can increase the accuracy of this result. 7.6.3 Other results Other results coming from the model created by this study are: Annual expected fatalities (AEF): average number of fatalities per year for a municipality Annual expected econ. loss (AEL): average direct cost per year for a municipality Dimitris Detsis 7-15

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 7.7 Modelling mitigations 7.7.1 Methodology Chapter 5 referred to different seismic risk mitigation strategies; here a more complete tool for decision-making mitigations is going to be modelled. Building retrofit is the main mitigation strategy examined. In addition, influence of de-urbanization is examined. The term de-urbanization is used here to refer to policies of reducing building stock and population of an urban complex. This project compares seismic risk metrics for existing stock and future retrofitted stock. This is done by simply re running the methodology described in the previous paragraphs, with differentiated vulnerability curves for the building. This new vulnerability is actually the demanding performance of buildings after retrofit. 7.7.2 Retrofit programs and retrofit levels Three programs of retrofitting building stock are going to be examined. These retrofit programs are differentiated by their impact to the vulnerability of the building stock. As a proposal, these three retrofit programs comprehend a light, a moderate, and a full version. The light version includes all masonry buildings retrofitted to performance level 3, the moderate includes all masonry buildings and buildings constructed before 1985, retrofitted to performance level 2, and the full all buildings except those constructed after 1995 without a pilotis, retrofitted to level 1. retrofit program buildings to be strengthened retrofit level Light all masonry buildings level 3 Moderate all masonry and r/c buildings before 1985 level 2 Full all stock except r/c buildings after 1995 level 1 Table 7-6 retrofit programs The introduced retrofit levels are chosen based on an attempt to extend the philosophy of performance-based design of the US seismic codes, Figure 7-9 (FEMA-389, 2004) and the codes for retrofit, such as KANEPE (EPPO, 2009). Therefore, for a specific seismic scenario a certain capacity/damage is demanded by the structure, Figure 7-10. The levels, currently examined, are indicative and aim to show the methodology. Levels need to be calibrated together with limits for societal risk or cost benefit analysis for societal economic risk. The same applies for retrofit programs. Figure 7-9 shows graph relating frequency and expected performance measured in four levels: operational, immediate occupancy, life safety and near collapse. With the same approach but trying to use expected damage degree rather than the above characterization, three different performance targets are proposed by this study for the three different retrofit levels (Figure 7-10). 7-16 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Figure 7-9 building performance levels in FEMA-389 and BSSC 2001 (FEMA-389, 2004) Figure 7-10 building performance levels chosen for this study Figure 7-11 retrofit levels and vulnerability of existing typologies Dimitris Detsis 7-17

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering In Figure 7-11 retrofit levels (black lines) are compared to the vulnerability of different typologies (colour lines). It can be noted that level 3, has an effect only on masonry buildings, while level 2 has an effect also to r/c buildings before 1985. Level 1 suggests improvement for all buildings except those built after 1995 without a pilotis. 7.7.3 Cost benefit analysis Along with the results described on this paragraph a cost benefit analysis is performed, solely considering direct economic loss, to estimate maximum investment cost for retrofit (MAX RETROFIT COST=C). The meaning of this value is: «Maximum amount of money spend in retrofit in order the investment to have an NPV=0 in T=20 years, interest rate r=6%» C T t = AEL existing AEL 1 (1 + r) t retrofit The essence of this calculation is to give an indication on the cost that would be feasible to be spent for retrofitting in the 3 pre-mentioned levels of retrofit. If costs are calculated then someone can decide upon the level of retrofit. Within the frames of this study, an indication is given, in order to show that an FL curve can also support cost benefit analysis modelling. 7.7.4 Risk criteria Finally, regarding the societal risk limits, lines on the FN graph, can be created for Greek society and for every municipality. Three indicative lines are proposed in relation to the population of municipality. Similarly, for individual risk a limit should be set. These issues are discussed in a separate paragraph 9.6. 7-18 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 7.8 Conclusions 1. This study provides a risk estimation model in excel (Appendix A). This model can be used to estimate societal & individual risk (conceptual) for existing building stock of municipalities of Greece. Then similar direct economic risk is estimated for various seismic scenarios. Moreover, mitigation strategies (mainly building retrofit) are modelled in order to quantify the potential of these strategies and compare their effectiveness. 2. Similarly to the methodology on estimating societal risk metrics for floods, to formulate FN or FL curve for seismic risk theoretically probability density functions for earthquake event, damage degree of building and fatalities or direct economic loss are required. However, this study models the calculation using deterministic functions for damage degree (vulnerability curves) and for fatalities / direct economic loss. 3. Data for exceedance probability of peak ground acceleration for seismic different zones, vulnerability of buildings in Greece, fatalities rate & direct cost in relation with damage degree are acquired from Greek and international bibliography. Inventory data is available from Greek National Statistical Authority. The area of the buildings is not available. To estimate average building area accurately enough, urban coverage factor of the city was estimated qualitative (i.e. for Athens is 50%) 4. For a municipality of Greece an FN-curve (societal risk) and an FL-curve (societal economic risk) are formulated. At the same time, two conceptual individual risk metrics are given, one averaged and one maximum, each one with a different role. The utility of the proposed metrics will be discussed in chapter.9, before sensitivity analysis of parameters is elaborated (chapter.8) 5. Performance targets are proposed, inspired by the performance-based design issued by US seismic codes and Greek Code of Interventions. Three retrofit programs of the stock are examined indicatively (light, moderate and full). In each one different typologies are retrofitted and in different performance levels. 6. The model of this study is built in Excel; please find Input / Results Board and rest of the data in Appendix A. The printout of this report refers to the municipality of Athens. Finally, please also note that the best available sources and information have been used in preparing this report and these sources and information has been interpreted exercising all reasonable skill and care. Nevertheless, neither author nor reviewers accept any liability, whether direct, indirect or consequential, arising out of the provision of information in this report. Dimitris Detsis 7-19

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 7.9 References [7.1] ANTYK, Theodorakis S. and Vougioukas E. Priority policy for seismic retrofit [Report] / Seismic Retrofit of Existing Building (ANTYK). - Athens : Technical Chamber of Greece, 2001. - GREEK. [7.2] Coburn A. and Spence R. Earthquake Protection: John Wiley & Sons Ltd, 2002. - 2nd. [7.3] Elenas A Athens Earthquake Of 7 September 1999: Intensity Measures And Observed Damages / ISET Journal of Earthquake Technology. - March 2003. - 1 : Vol. 40. - pp. 77-97. [7.4] EPANTYK Pre-seismic Retrofit of existing buildings [Book] / ed. Buildings National Program for Seismic Retrofit of Existing. - [s.l.] : Technical Chamber of Greece (T.C.G.), 2007. - GREEK. [7.5] EPPO Greek Seismic Code / ed. (OASP) Earthquake Planning and Protection Organization. - Athens : Ministry of Environment, Planning and Public Works., 2003. - BUILDING CODE - GREEK. [7.6] EPPO Code of Interventions (KANEPE) [Book] / ed. OASP Earthquake Planning and Protection Organization: Ministry of Environment, Planning and Public Works, 2009. - GREEK - BUILDING CODE. [7.7] European Standard Eurocode 8: Design of structures for earthquake resistance - Part1: General rules, seismic actions and rules for building. - 2003. - pren 1998-1. [7.8] FEMA-389 Primer for Design Professionals Communicating with Owners and Managers of New Buildings on Earthquake Risk: Federal Emergency Management Agency, 2004. [7.9] Grossi P. Modeling Seismic Mitigation Strategies / Risk Assessment, Modeling and Decision Support / book auth. Bostrom A, French S and Gottlieb S.: Springer Berlin Heidelberg, 2008. - Vol. 14. - 10.1007/978-3-540-71158-2. [7.10] Kappos A. J. and Dimitrakopoulos E. G. Feasibility of pre-earthquake strengthening of buildings based on cost-benefit and life-cycle cost analysis, with the aid of fragility curves / Natural Hazards. - August 21, 2007. - Vol. 45. - pp. 33-54. - DOI 10.1007/s11069-007-9155-9. [7.11] Kappos A.J. [et al.] Correlation of structural damage of R/C buildings with economic loss: calibration based on earthquakes of Athens 1999 and Thessaloniki 1978 / 15th Concrete Conference. - Alexandroupoli : TGC-ETEK, 2006. [7.12] Papaioanou Ch. and Voulgaris N et al The Utilization of New Seismological Data in the Compilation of the New Seismic Hazard Map of Greece / 3th National Conference of Seismic Mechanics & Technical Seismology. - 2008. - GREEK. [7.13] Seligson H. A., Shoaf K. I. and Kano M. Development Of Casualty Models For Non-Ductile Concrete Frame Structures For Use In Peer s Performance-Based Earthquake Engineering Framework / 8th U.S. National Conference on Earthquake Engineering. - San Francisco, California, USA : / 2006. - Vol. No. 917. [7.14] Solomos G., Pinto A. and Dimova S. A Review Of The Seismic Hazard Zonation In National Building Codes In The Context Of Eurocode 8: Support to the implementation, harmonization and further development of the Eurocodes. - Brussels : JRC - European Commission, 2008. - EUR 23563 EN - 2008 [7.15] Greek National Statistical Authority [Online]. - http://www.statistics.gr. 7-20 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report.8 Sensitivity analysis Mitigation effectiveness This chapter presents conclusions of sensitivity analysis. Thus, it indicates the most critical parameters of the estimation model developed during master thesis and recommends directions for future research. 8.1 Sensitivity analysis 8.1.1 How it is done Sensitivity analysis aims to recognize critical assumptions, compare alternate model structure, guide future data collection, and detect important criteria. Moreover, it ensures model s accuracy and simplifies sometimes guides to simplifications in methodology. Sensitivity analysis was actually a continuous process during the process of constructing this model. This paragraph explains how main assumptions were made during the process and presents influence of model s parameters to the result. Sensitivity analysis is done by changing values in certain parameters of the model while keep the others stable. In the following comparative results are given for most of the parameters of the model. Inventory data concern municipality of Athens. Results are also gathered in appendix B. 8.1.2 Hazard This project, in an effort to be synchronized with recent research about Probabilistic Seismic Hazard Assessment for regions of Greece examined several references (references). However, they were not in such point to be used for the model for the following reasons: they were not providing understandable result, exceedance probability curves for different period are differentiate very little, they are focusing on specific regions Differences between FN curve for hazard curve of equation 7-1 and this proposed by other references (see appendix C) are focalizing on low frequency earthquakes. The result is significantly different there, guiding to the conclusion that further research is required to specify accurately hazard intensity in such low frequent events. See Appendix B for figures. In general, if exceedance probability of PGA (hazard) increases, as expected, exceedance probability of fatalities and direct economic loss increase too. This means that FN curve and FL curve shift vertically (up when hazard increases and down when hazard decrease). Please refer to graphs 1.1 and 1.2 in appendix B. Dimitris Detsis 8-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 8.1.3 Vulnerability Like hazard, vulnerability parameter has the similar influence to risk criteria. Graphs 1.3 & 1.4, show FN curve and FL curve that swift vertically. It is notable the same behaviour of FN curve in case of floods (see paragraph 4.4.5 ) Hazard and vulnerability coupled can give probability of damage. If probability of damage decreases then FN curve swifts vertically down, similar with FN curve for floods in The Netherlands. 8.1.4 Mortality function Criticality of coefficients of mortality function is also examined in the frames of sensitivity analysis. When parameters described in paragraph 7.3.3 are increased FN curve swifts horizontally to the left, while when they decrease curve swifts to the right. Graphs 2.1, 2.2 and 2.3 in the appendix B show that. Furthermore regarding the most critical parameter from those three, previous graphs together with graph 2.4, indicate M3: occupants trapped. 8.1.5 Average area of individual building As mentioned in paragraph 7.4.2 average area of buildings in the region is the only missing parameter from inventory data. Graph 3.1 on appendix B shows the influence of changing this parameter to FL-curve. Contrary as already mentioned average area of building does not influence fatalities. 8.1.6 Calibration of cost function To calibrate cost function also influence on the FL curve was examined. In the following graph, three curves represent three calculations for different cost functions: 1. Original one from reference paper (Kappos, et al., 2006) 2. Original one forced to take zero value for damages lower than 1% 3. Proposal of this study equation 7-6 The third was chosen as it ends up giving result that is more reasonable. This is based on similar curves from reference (Grossi, 2008) and on explanation given on paragraph 7.3.4 8-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Figure 8-1 cost function selection 8.2 Effectiveness of mitigation strategies 8.2.1 Parameters examined Besides sensitivity analysis, other parameters rather than hazard, vulnerability, mortality coefficients and average building area have been examined for their influence on seismic risk metrics. Even if these parameters are fixed values for a municipality, it is interesting to examine their influence. The following table lists all the parameters examined here: parameter mitigation strategy 1 hazard seismic isolation 2 vulnerability retrofit programs 3 nr of buildings remove buildings 4 population lower population 5 de-urbanism remove buildings + population Table 8-1 - model parameters and mitigation strategies 8.2.2 Results By lowering the value of each parameter by 20%, the result on FN and FL curve can be observed in Figure 8-2 & Figure 8-3 respectively. These results depends also on the municipality, it is reminded that the case of municipality of Athens is examined here. However, in order to make conclusions it is necessary to see also the influence on AEF (annual expected fatalities) and AEL (annual expected loss) when these parameters changing, in Figure 8-4 and Figure 8-5. Dimitris Detsis 8-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Conclusion and recommendations can be found in the following paragraph. Figure 8-2 influence of migration strategies on FN curve Figure 8-3 influence of migration strategies on FL-curve 8-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Figure 8-4 - influence of mitigation strategies in AEF Figure 8-5 influence of mitigation strategies in AEL Dimitris Detsis 8-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 8.3 Conclusions & Recommendations Conclusions and recommendations derived from sensitivity analysis (paragraph 8.1) and from investigation of effectiveness of main mitigation, strategies (paragraph 8.2) are gathered here. 1. Increasing / decreasing hazard or vulnerability causes FN and FL curve to shift up / down vertically. If fatalities rate increases / decreases FN curve, will shifts horizontally right / left. Horizontal shift is the influence of average building area in FL curve. This behaviour of the societal risk metrics is similar with the case of flood risk. If probability of damage changes curve moves vertically, while if probability of consequences changes curve moves horizontally. 2. This result proves the ability of FN and FL curve (and societal risk metrics) as tools of decision making, whereas are able to depict the influence of different parameters in one graph. Someone can judge the effectiveness of mitigation program when he checks those two graphs. 3. Average building area has no influence on FN curve. This is a very important remark since average building is now the most unreliable value. This derives actually from the mathematics of the model. However, because influence of the value of average building area is considerable for direct economic loss, it is recommended values used in this study to be validated. 4. Criticality of mortality function is low. Even if a detailed research over the relation between damage and fatalities is going to be very interesting, it should not be a priority. On the other hand, the most critical parameters are hazard and vulnerability. Regarding hazard, it is more a matter of researcher and scientists to give trustworthy results. Vulnerability of buildings is also a critical parameter. Vulnerability (or derivation of it damage probability matrices / fragility curves) is necessary to be investigated continuously as it is also a parameter influenced by the time (as the stock grows old it becomes more vulnerable due to deterioration). However, for model suitable for decision-making in top level or medium level, which sets priorities, vulnerability curves presented by EPANTYK (2007) are appropriate. 5. Concerning mitigation strategies, graphs of paragraph 8.2 can give indication of which is the most effective strategy. In order to answer which is the most effective strategy cost parameters must be also known. However, some conclusions can be made without knowing costs. Removing/demolishing most vulnerable buildings need to be integrated with decreasing population (de-urbanization) in order to be effective. Since de-urbanization is a time consuming, costly while not so effective measure and seismic isolation is not usually an option for existing buildings, retrofit is the most effective way of mitigating risk (retrofit here has the meaning of strengthening or rebuild). 8-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report.9 Potential use of metrics in safety policy Problems of seismic risk estimation referring to seismic risk mitigation discussed on chapter.1 Translating what was presented in chapter.3 and.4, chapter.6 proposed the use of societal & individual risk metrics in the frames of possible government intervention. Then chapter.7 depicted how risk metrics, recently used in flood risk safety policy, can be estimated in the case of Greek municipalities. This chapter demonstrates how this metrics can be useful for decision-making. The first part describes main seismic risk management steps and positions the metrics, while the second part examines a specific example implementing the proposed procedure. The third part deals with recommendations for improvements (discussion about limits and indirect loss) 9.1 Seismic risk management steps 9.1.1 Decision-making flow chart As mentioned, whatever the government intervention, risk estimation is vital to be able to set priorities and decide whether buildings, municipalities, or regions are safe enough. In chapter.6 a possible scheme for decision-making was discussed. A part of it is depicted in Figure 9-1. Figure 9-1 decision making structure Dimitris Detsis 9-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 9.1.2 Position of the proposed metrics Figure 9-1 also shows the positions of societal risk and individual risk as estimated by the developed model. In more detail, this study proposes the following: For Top level - Central administration decision making 1. Graph with societal risk (FN curve) of all municipalities (& total) 2. Graph with societal economic risk (FL curve) of all municipalities (& total) 3. Ranking for average & maximum individual risk for all municipalities (& total) 4. Ranking of annual expected loss per capita for all municipalities (& total) It is noted that the total seismic risk for the country is not a subject of this thesis. To calculate total seismic risk for the whole of the country, correlation effects (dependencies) of earthquake events have to be considered. However, it is recommended such an estimation to be implemented in the future, since it gives an overview and helps monitor safety plan implementation. For Medium level - Local administration decision making 1. Graph with societal risk (FN curve) of all typologies and total 2. Graph with societal economic risk (FL curve) of all typologies and total 3. Graph with societal risk (FN curve) for 3 retrofit programs 4. Graph with societal economic risk (FL curve) for 3 retrofit programs 5. Graphs with influence of retrofit programs to individual risk and AEL/capita The decision is based on risk metrics for every building typology and for the total. This project proposed the following options for retrofit programs: no retrofit (other measures with minor effectiveness) light (all masonry building in retrofit level 3) moderate (all masonry and all buildings built before 1985 level 2) full (all stock except new buildings built after 1995 level 1) These are indicative retrofit programs (see paragraph 7.7.2 ) The goals of a safety plan of the top or medium level can be: reduce risk to areas where it is higher than others (comparing & prioritizing), reduce risk as low as possible or according to social accepted criteria / limits The first point is possible within the developed methodology of this project, and will be demonstrated in the next paragraph by analysing two case studies. Risk acceptance criteria / limits are only discussed in paragraph 9.6. Finally, the model developed in this study is not appropriate for making conclusions on individual building cases. More analytical models and detailed modelling are necessary. Instruments and strategies for this scale are discussed in chapter.3 and paragraph 5.3.2 (analytical methods, performance-based design, cost benefit analysis of the owner etc). 9-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 9.2 Case study 1 - Top level decision making 9.2.1 Description of the example The four largest cities of Greece concern the example presented here. The table below gives basic social-economic characteristics and seismic zone information. The graphs on the table show the distribution of building inventory for the 4 cities. avg build occupancy seismic municipality population buildings area area rate zone Athina 800K 63K 39 s.km. 1000 79 I Thes/niki 390K 24K 18 s.km. 1250 76 I Pireas 180K 27K 11 s.km. 750 112 I Patra 170K 31K 57 s.km. 750 137 II Table 9-1 characteristic of cities Athina (Athens) Thessaloniki Pireas Table 9-2 inventory of the four cities Patra 9.2.2 Results The following graphs are the results of the developed estimation model: Figure 9-2 - individual risk & individual economic risk Dimitris Detsis 9-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Figure 9-3 - graph with societal risk (FN curve) of municipalities Figure 9-4 - graph with societal economic risk (FL curve) of all municipalities 9-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 9.2.3 Analysis The city of Athens has the larger societal risk both for life and for economic wealth. However, Patra has the largest individual risks for average values (life and wealth). In Athens and Thessaloniki due to larger population, high societal risk is distributed in more people thus averaged individual metrics score lower. Maximum individual exposures are similar for the four cases. The crossing of curves of societal economic risk of Thessaloniki and Patra is also notable, suggesting that Patrai (curve more to the left in the high frequencies) needs faster measures than Thessaloniki. Another interesting remark is that Athens has a considerably higher risk and 4 fatalities for relatively frequent events. 9.2.4 Decision (proposal) Consequently, looking into the above results, central government should plan its intervention focusing on the followings. 1. Athens needs immediate measures to avoid fatalities with exceedance probability 50% in 1 year. In general Athens should absorb more of the resources since presents higher risk in societal metrics 2. Patra must as well be in the centre of interest, since it has high scores on the average individual metrics and, as it can be seen on the societal economic risk curve, it has relatively high losses for frequent events. 3. Thessaloniki and Pireas will share the rest of the fictitious resources, since the first scores higher in societal metrics while the latter scores higher in individual metrics. 4. There are no disproportional exposures between the 4 cities (with deliberation) At this point, it is important to note that the decision might change if someone had an overview of indirect losses too. Moreover, it is still difficult to make strong conclusions based on risk-risk comparison, since the costs of risk reduction might vary drastically from case to case. Analytical results can be found in appendix A. 9.3 Case study 2 - Medium level decision making 9.3.1 Description of the example From the proposals of the previous case study, the policy maker of Athens municipality would have received guides to reduce the seismic risk in the municipality. Then his choice would be probably to retrofit (renew) the building stock. Dimitris Detsis 9-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 9.3.2 Results The following graphs are the results of the developed model (see also pages 9-6 & 9-7) Figure 9-5 individual risk & individual economic risk for existing and retrofitted stock 9.3.3 Analysis The main problem of Athens is the large and vulnerable stock of buildings built before 1985 (masonry and r/c). Masonry buildings increase exceedance probability of 4 fatalities to to50% per year. Even light retrofit program 1 would marginally improve the situation, while it would decrease consequences in events of exceedance probability higher than 50% in 50 years. The main difference between moderate and full retrofit is in events with low annual frequency. 9.3.4 Decision (proposal) Consequently, looking into the above results local government should plan its intervention focusing on the followings. 1. Immediate actions are necessary in order to decrease fatalities and economic losses of frequent events. 2. Retrofit program 3 light (retrofit masonry) can reduce risk for those events and decrease scores of individual metrics 3. If Retrofit program 2 moderate follows then there will be improvement. 4. Masonry buildings are areas with disproportional individual exposures as it is derived by the comparison of averaged and maximum individual risk. At this point, it is important to note that the decision might change if someone had an overview of indirect losses too. Moreover, there is no consideration of retrofit costs. Analytical results can be found in appendix A. 9-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report Figure 9-6 - graph with societal risk (FN curve) of all typologies and total Figure 9-7 - graph with societal economic risk (FL curve) of all typologies and total Dimitris Detsis 9-7

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Figure 9-8 - graph with societal risk (FN curve) for 3 retrofit programs Figure 9-9 - graph with societal economic risk (FL curve) for 3 retrofit programs 9-8 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 9.4 Evaluation of the proposed model This thesis aim is not to present definite seismic risk results, but to demonstrate the potential of using societal risk & individual metrics as implemented in human-hazard cases and recently in Dutch flood safety policy, to seismic risk mitigation policy in Greece. Therefore, the result of this thesis will mainly focus on the evaluation of the methodology rather than the results. The previous paragraphs presented an example of how the proposed prototype model, can be used for decision making regarding seismic risk mitigation. The first conclusion is that the model can help central and local administration to set priorities for mitigation between municipalities and retrofit programs referring to structural typologies of buildings. The case studies show that risk metrics and the model quantify them can be useful tools for deciding: Which municipality should absorb more resources? Is it urgent or can it be planned for the future? Which mitigation strategy is more efficient? Which is the influence of a certain retrofit program? Besides there are important benefits from the described methodology like: Seismic risk is quantified for all the scenarios of earthquakes. An FN-curve as an outcome gives a total overview. In that way knowledge is gained for the consequences of earthquake events with lower or larger magnitude than those addressed by building codes. Events with high numbers of fatalities, which can cause disruption to the whole of the country and disproportional individual exposures, can be targeted directly. The first can be viewed from societal risk results (FN-curve) and the later in individual risk contours, which can be drawn after collecting information on the location of each building. The controversial issue of human life monetization can be avoided. Fatalities (deaths or even injuries) are distinguished from cost-benefit analysis and there is no need give a price to them in order to make a decision. The decision maker, (possibly not an earthquake expert), only deals with probabilities and consequences, while having a general visualization of the effectiveness of alternative mitigation strategies. In addition, since loss is associated with return period (annual exceedance frequency), resources can be distributed more efficiently in time. Finally, the proposed methodology gives the opportunity of monitoring the progress of safety plan and it is scalable for central and local administration. Besides providing important input, it initiates the discussion for the institution of societal & individual acceptance criteria (further discussion on par. 9.6 Dimitris Detsis 9-9

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 9.5 Possible improvements / recommendations for research 9.5.1 Regarding risk estimation components Hazard assessment The estimation model of this study uses data for exceedance probability of PGA from a research conducted in 1993. Updated probabilistic seismic hazard assessment would improve the accuracy of the result. These kind of studies should focus not only on determining PGA (or other motion parameters) associated with reference return period (475years) but also on providing curves for a wider range (return periods of 1 year to 10000 years). Vulnerability data Vulnerability data provided by the Greek research programs is useful and approachable. Moreover, it provides information for the majority of buildings, while the relation with the inventory data makes it easier to handle. The level of accuracy is good enough for the scale of decision making on top and medium level. Introduction of probabilistic functions for vulnerability (fragility curves) is also an option. However, it will make the estimation more complicated. Finally, the idea of creating a large database with analytical vulnerability data cannot be completed earlier than an estimated period of 10 years. Actually, it needs to be prioritized using more rough risk estimator models like the one proposed. Loss functions It should be reminded that in the model all the buildings of a municipality are considered of equal size, with the same occupancy and classification. Thus, a first improvement for the proposed model could be introducing factors related to those 3 parameters. Equally to vulnerability, these parameters can be stored in a database and help for the computerization of the estimation procedure. Furthermore, data regarding the size of the buildings (floor area) which is not provided by the Greek statistical service should be provided (also request of Greek researchers). Given those parameters, the results of the proposed model would be much more trustworthy. Finally, as already mentioned, the mortality function should be more accurate and related to Greek building characteristics. Indirect loss Risk estimators of this study did not take into account indirect losses. As discussed for both the case of floods and earthquakes indirect loss and quantification of the intangibles is difficult. An idea could be to try to relate them with direct loss (factored or exponential). However, since indirect losses mainly have a non-technical nature, this research proposes to evaluate them qualitatively and place them in a broader multi-criteria evaluation decisionmaking, together with the results of expected direct losses. 9-10 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 9.5.2 Regarding methodology & results of the model Correlation The calculation methodology of the proposed model is semi-probabilistic. Loads on different buildings are assumed to be perfectly correlated (depended); the model is based on the assumption that when ground shakes, every building is subjected to the same peak ground acceleration (PGA). Regarding strength, vulnerability relates damage degree to PGA. In the model, it is assumed that every building of the same typology is always damaged to the same extent. This implies that building strengths are also assumed to be perfectly correlated. The effect of this assumption should be investigated. Individual risk The proposed model estimates individual risk and individual economic risk. It should be reminded that individual risk is defined as the probability of death of an unprotected person in a location. However, the calculated individual risk metrics deviate a little from this definition for two reasons: there were no data available about the position of the buildings inside the municipality, such an estimation requires more analytical seismic risk estimation building by building Thus, individual risk metric is broken down into two sub metrics, an average and a maximum individual risk (see par.7.6.2 ). These two can give an indication and guide for further assessment, possibly towards the bottom level (owner). Anyway, it should be clear by now that the methodology, proposed in this study, is not appropriate to spot out cases of single buildings, which have high risk due to parameters different from those examined (collateral / indirect seismic effects, and/or illegalities). However, these buildings must be spotted and measures have to be taken, (in the case of 1999 Athens earthquake, 39 out of 143 fatalities were caused by the collapse of a single factory building) 9.5.3 Next steps Enhance mitigation modelling In This project, after risk estimation of existing situation, two mitigation alternatives were modelled (retrofit of buildings and de-urbanization). The model is appropriate for evaluating effectiveness of mitigation. Thus, it is recommended that more alternatives should be compared (i.e. insurance policies). Moreover, retrofit programs, which were indicatively proposed, must be calibrated according to available resources, knowledge and technology. Cost modelling of mitigation (retrofit) One very essential step forward would include estimation of mitigation alternative costs and main retrofit program costs. Awareness of cost of risk reduction is necessary for the debate about appropriate (efficient/feasible) societal and individual risk acceptance criteria. Dimitris Detsis 9-11

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 9.6 Societal & Individual risk limits Introducing limits is discussed in this paragraph. Regarding a decision maker, societal risk or individual risk metrics does not mean anything, if they cannot be compare either to another known situation or to limits. As mentioned, to formulate these limits, social parameters must be examined and this is out of the field of this study. However, a proposal is referred to, in order to initiate a discussion about enactment of such limits, acceptance criteria will go together with the proposed risk metrics and methodology. The least that can be mentioned is that risk acceptance criteria are not nature laws. They are based on subjective judgements of the society or sometimes on the impression decision makers have on the society judgement and the way it prioritizes several activities and their risks. A societal risk criterion can be a line (or curve) on the FN graph. If the FN-curve of the hazard is below this line, risk is acceptable. Vrijling, et al., (1998) discussed among other hazards an FN criterion for floods. Two indications were given depending on the considered factor of voluntariness β. See in Figure 9-10 line I β=0,1 and red dotted line β=1. In this study, which compares municipalities with wide range of population, besides voluntariness, this parameter should be considered. One fatality is more noticeable in a municipality of 1000 people than in one with 500000 people. Therefore, different limits depending also on the population are discussed at this point. Of course, they are also indicative, since this kind of criteria are set according to the society wishes and the difficulty to be achieved. Using the limit proposed by (Vrijling, et al., 1998) for voluntariness β=0,1 as a limit for an area with a population of 10k-50k limit I. Limit II is proposed for municipalities of 50k-500k inhabitants is represented as the line that connects the top of Vrijling limit for β=0,1 and β=1. For larger municipalities limit II is doubled to form line III, see following Table 9-3 and Figure 9-10. Using acceptance criteria, the decision may become easier. For example for Athens, Figure 9-11, the decision maker would have decided to organize retrofit program 2. However, even in this case if retrofit costs were insufferable for the community the decision would not be accepted. Similar risk acceptance criteria / limits can be set for individual risk. 9-12 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report β=0,1 combination I ΙI ΙΙI 1,0E-05 180 1600 3200 4,0E-04 30 120 240 2,1E-03 13 40 80 1,4E-02 5 10 20 3,4E-02 3,25 5 10 Table 9-3 societal risk limits Figure 9-10 proposal for societal risk limits Figure 9-11 - Athens existing & retrofitted stock compared with limits Dimitris Detsis 9-13

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 9.7 Conclusions 1. Main outputs of the proposed model: graphs with societal (& economic) risk of different municipalities, ranking of individual metrics can support top level central administration intervention. Graphs with societal (& economic) risk of different building typologies inside a municipality and values of individual risk metrics for existing and retrofitted stock can support medium level local administration intervention. The proposed graphs as outputs of the developed model guide the decision maker at least for priorities policy. 2. Two case studies are presented that show the use of the aforementioned risk metrics for different levels of government decision-making. The first simulates top level (central government) decision-making, setting priorities for retrofit between municipalities, and the second simulates medium level (local government) decision-making, setting priorities for a retrofit program amongst different structural typologies of buildings. The case studies show that the risk metrics and the model to quantify them can form useful tools for deciding which municipality should absorb more resources, whether mitigation is urgent, which mitigation strategy is most efficient, and how alternative retrofit programs influence risk levels. 3. There are important benefits from the implementation of the described methodology. Firstly, the decision maker only deals with probabilities and consequences and has a general overview, thus he/she may distribute resources and time in a more (cost) effective way. Moreover, human life is distinguished from cost-benefit analysis (no monetization). Events with high numbers of fatalities, which can cause disruption to the whole of the country as well as disproportional individual exposures, can be targeted directly. Finally, it gives the opportunity to monitor the progress of a safety plan, and is scalable for central and local administration. 4. Of course, the model is only a prototype. Further refinements are advised regarding input parameters, methodology and results. 5. The most important direction for further research refers to the costs of alternative retrofit programs or general mitigations strategies, which are necessary for the debate about appropriate (efficient/feasible) societal & individual risk acceptance criteria. 9-14 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report PART IV - CONCLUSIONS.10 Conclusions The most important conclusions from all chapters are revised at this point. The first paragraph outlines the main conclusions (answers to research questions); the second one condenses the proposal while the last one discusses main recommendations for further research. 10.1 Main conclusions (answers to research questions) 1. Earthquake and floods threaten Greece and The Netherlands respectively. Although the technical part of hazard is different, the cornerstones of the hazard risk management do not differ. The double nature of risk must be noted technical and non-technical as well as the different perspective of different individuals or society. Government intervention should be taking into account in both aforementioned perspectives. 2. The safety assessment of an existing structure differs from a new one since, to increase safety a lot more need to be spend, the remaining lifetime is less than the standard period of 50 years and finally performance measurement can be made to reduce uncertainty. 3. Individual and societal risk metrics are assessing the impact of flood in numbers of human losses. The first is the probability of death for an individual in a certain location and the second relates the exceedance probability of an event and potential number of fatalities. The graphical representation of societal risk is an FN-curve while individual risk is a single value. 4. Recent developments in Greece regarding seismic safety mostly aim at vulnerability and risk evaluation, the publishing of a technical building Code of Interventions, and mapping out a seismic safety plan for the country. Meanwhile, experts ask for government intervention proposing organisational change and distribution of roles / liabilities among different clusters. 5. In the case of the Netherlands, due to the nature of the flood hazard and protection scheme (public flood defences), the government is strongly involved in flood risk mitigation. Despite the differences between the protection schemes for large-scale floods (strengthening dikes rather than protecting buildings) and earthquakes (strengthening buildings), this project proposes the translation (=act of converting) of individual and societal risk metrics to the case of seismic risk in Greece. 6. The prototype model, worked out during this thesis, demonstrates that risk metrics of societal & individual risk can be estimated by using probabilistic function for hazard and deterministic transfer functions for damage along with expected losses. The available data is enough for a first rough calculation and testing of the methodology. Since economic losses of earthquakes are also critical, the risk metrics were also converted, to estimate economic risk. Dimitris Detsis 10-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering 7. The outcome is an instrument, which provides information for decision making, to central and/or to local authorities. According to it, the decision maker can set priorities or decide whether a region is safe enough, based on limits/risk acceptance criteria. 10.2 The proposal Based on research on the cornerstones of flood risk management practices in The Netherlands, and on testing on the application in seismic risk estimation in the case of Greece this master thesis proposes to Greek authorities: To utilize estimation of societal & individual risk metrics, for fatalities and economic loss, to support two levels (central & local) of government decision-making concerning seismic risk mitigation. The main benefits would be: 1. overview and comparison of alternative mitigation strategies 2. effective distribution of resources and time 3. human life is distinguished from cost-benefit analysis (no monetization) 4. events with high numbers of fatalities, which can cause disruption to the whole of the country, disproportional individual exposures, can be targeted directly 5. the decision maker only deals with probabilities and consequences 6. the opportunity to monitor the progress of a safety plan This master thesis also demonstrates the methodology for the quantification of aforementioned metrics and provides a prototype an estimator model. 10.3 Recommendations for research Finally, among several directions for further research the most important are: regarding model s input parameters introduce, especially for local administration, more social characteristics of buildings like value, importance etc and more accurate approximation of their surface keep hazard, vulnerability and loss function updated regarding model s calculation methodology research on the effect of assumed perfect correlation between loads and strengths improve the methodology of individual risk metrics calculation regarding next steps for improved decision making enhance retrofit modelling and model cost of estimation of mitigation alternatives discussion about levels of expected indirect losses initiation of the debate about appropriate (efficient/feasible) societal & individual risk acceptance criteria._ 10-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report 10.4 References (ALL) Applied technology Council ATC-40 - Seismic evaluation and retrofit of concrete builidngs. - 1996. - Vol. I. Brinkhuis-Jak M. [et al.] Cost benefit analysis and flood mitigation in the Netherlands [Journal] // Heron. - 2004. - 1 : Vol. 49. Burton I, Kates R.W. and White G.F. The Environment as Hazard [Journal]. - New York : Oxford University Press. Coburn A. and Spence R. Earthquake Protection [Book]. - [s.l.] : John Wiley & Sons Ltd, 2002. - 2nd. Diamantidis Dimitris and Bazzurro Paolo Safety acceptance criteria for existing structures [Journal] // Workshop on Risk Acceptance for existing structures. - Stanford University, USA : [s.n.], March 2007. Eijgenraam C.J.J. Optimal safety standards for dike-ring areas [Report] : Discussion Paper. - The Hague : CPB Netherlands Bureau for Economic Policy Analysis, 2006. Elenas A ATHENS EARTHQUAKE OF 7 SEPTEMBER 1999: INTENSITY MEASURES AND OBSERVED DAMAGES [Journal] // ISET Journal of Earthquake Technology. - March 2003. - 1 : Vol. 40. - pp. 77-97. European Standard Eurocode 8: Design of structures for earthquake resistance - Part 3: Strengthening and repair of buildings. [Book]. - 2003. - pren 1998-1. European Standard Eurocode 8: Design of structures for earthquake resistance - Part1: General rules, seismic actions and rules for building [Book]. - 2003. - pren 1998-1. FEMA-356 Prestandard & Commentary for the Seismic Rehabilitation of Buildings [Book]. - Washington DC : Federal Emergency Management Agency, 2000. FEMA-389 Primer for Design Professionals Communicating with Owners and Managers of New Buildings on Earthquake Risk [Book]. - [s.l.] : Federal Emergency Management Agency, 2004. Giardini D., Jimenez M.J. and Grunthal G. European-Mediterranean Seismic Hazard Map [Report]. - [s.l.] : European Seismological Commission, 2003. Grossi P. Modeling Seismic Mitigation Strategies [Book Section] // Risk Assessment, Modeling and Decision Support / book auth. Bostrom A, French S and Gottlieb S. - [s.l.] : Springer Berlin Heidelberg, 2008. - Vol. 14. - 10.1007/978-3-540-71158-2. Jongejan R. B., Jonkman S. N. and Maaskant B. The potential use of individual and societal risk criteria within the Dutch flood safety policy (part 1): basic principles. [Conference] // Reliability, Risk and Safety, ESREL / ed. R.Bris C.G. Soares, S.Martorell. - 2009. Jongejan R.B. How safe is safe enough? The government s response to industrial and flood risks [Report] : Doctoral thesis / Delft University of Technology. - Delft, The Netherlands : [s.n.], 2008. Kappos A. J. and Dimitrakopoulos E. G. Feasibility of pre-earthquake strengthening of buildings based on cost-benefit and life-cycle cost analysis, with the aid of fragility curves [Journal] // Natural Hazards. - August 21, 2007. - Vol. 45. - pp. 33-54. - DOI 10.1007/s11069-007-9155-9. Dimitris Detsis 10-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Maaskant B., Jonkman S. N. and Jongejan R. B. The use of individual and societal risk criteria within the Dutch flood safety policy (part 2): estimation of the individual and societal risk for the dike rings in the Netherlands [Conference] // ESREL. - 2009. Pinto E.P. THE EUROCODE 8 PART 3: THE NEW EUROPEAN CODE FOR THE SEISMIC ASSESSMENT OF EXISTING STRUCTURES [Journal] // ASIAN JOURNAL OF CIVIL ENGINEERING (BUILDING AND HOUSING). - 2005. - 5 : Vol. 6. - pp. 447-456. Priemus H. Unification of the European building market: Possible consequences for the Dutch construction industry [Journal] // Housing and the Built Environment. - March 1991. - 1 : Vol. 6. - pp. 35-46. Seligson H. A., Shoaf K. I. and Kano M. DEVELOPMENT OF CASUALTY MODELS FOR NON-DUCTILE CONCRETE FRAME STRUCTURES FOR USE IN PEER S PERFORMANCE-BASED EARTHQUAKE ENGINEERING FRAMEWORK [Conference] // 8th U.S. National Conference on Earthquake Engineering. - San Francisco, California, USA : [s.n.], 2006. - Vol. No. 917. Solomos G., Pinto A. and Dimova S. A REVIEW OF THE SEISMIC HAZARD ZONATION IN NATIONAL BUILDING CODES IN THE CONTEXT OF EUROCODE 8 [Report] : Support to the implementation, harmonization and further development of the Eurocodes. - Brussels : JRC - European Commission, 2008. - EUR 23563 EN - 2008. T. P. Tassios, Kostas Syrmakezis, RC Moment Frame Building : Dual System - Frame with Shear Wall, GREECE / HOUSING REPORT / World Housing Encyclopaedia (http://www.worldhousing.net/ ) Tsapanos T.M., Mäntyniemi, P. & Kijko, A A probabilistic seismic hazard assessment for Greece and the surrounding region including site-specific considerations [Journal] // Annals of geophysics. - December 2004. - 6 : Vol. 47. - pp. 1675-1688. Unie van Waterschappen Climate Change and Dutch Water Manageme [Report]. - the Netherlands : Unie van Waterschappen, 2008. van Dantzig D. Economic Decision Problems for Flood Prevention [Journal] // Econometrica. - [s.l.] : The Econometric Society, July 1956. - 23 : Vol. 24. - pp. 276-287. Vlek C. and Keren G. Behavioural decision theory and environmental risk management: Assessment and resolution of four survival dilemmas [Journal]. - [s.l.] : Acta Psychologica, 1992. - Vol. 80. - pp. 249-278. Vrijling J.K., van Gelder P.H.A.J.M. and Ouwerkerk S.J. Criteria for acceptable risk in the Netherlands [Book Section] // Infrastructure Risk Management Processes: Natural, Accidental, and Deliberate Hazards / book auth. Taylor Craig E. and Vanmarcke Erik. - [s.l.] : ASCE, 2006. Vrijling J.K., van Hengel W., l and Houben R.J. Acceptable risk as a basis for design [Journal] // Reliability Engineering and System Safety. - [s.l.] : Elsevier, January 1998. - 1 : Vol. 59. - pp. 141-150. Vrouwenvelder A.C.W.M. and Scholten N.P.M. Veiligheidsbeoordeling bestaande bouw - Achtergrondrapport bij NEN8700 [Report] / Bouw ; TNO. - Delft : TNO Bouw, 2008. - p. 39. - 2008-D-R0015/B. Vrouwenvelder T. and Scholten N. Assessment criteria for Existing structures [Journal] // Structural Engineering International / ed. Editorial SEI. - November 17, 2010. - Vol. 1. Woditsch Richard; Plural- Public and Private spaces of the polykatoikia in Athens; Berlin; 2009 10-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report TU DELFT lecture notes CUR Probability in Civil Engineering [Lecture Notes]. - Delft : TUDelft, 2006. - Vol. 1. - (CT4130). Verhaeghe R.J. Plan & Project Evaluation [Lecture Notes]. Delft : TUDelft, 2007 GREEK ANTYK, Theodorakis S. and Vougioukas E. Priority policy for seismic retrofit [Report] / Seismic Retrofit of Existing Building (ANTYK). - Athens : Technical Chamber of Greece, 2001. - GREEK. EPANTYK Pre-seismic Retrofit of existing buildings [Book] / ed. Buildings National Program for Seismic Retrofit of Existing. - [s.l.] : Technical Chamber of Greece (T.C.G.), 2007. - GREEK. EPPO Greek Seismic Code [Book] / ed. (OASP) Earthquake Planning and Protection Organization. - Athens : Ministry of Environment, Planning and Public Works., 2003. - BUILDING CODE - GREEK. EPPO Code of Interventions (KANEPE) [Book] / ed. OASP Earthquake Planning and Protection Organization -. - [s.l.] : Ministry of Environment, Planning and Public Works, 2009. - GREEK - BUILDING CODE. EPPO Recommendations for pre-earthquake and after-earthquake intervention in buildings [Book] / ed. OASP Earthquake Planning and Protection Organization -. - Athens : Ministry of Environment, Planning and Public Works, 2001. - GREEK. Institute of Local Government Liabilities & Roles of Local Governments in Civil Protection for Mitigation of Natural Hazards [Report]. - Athens : Institute of Local Government, 2008. - GREEK. Kappos A.J. [et al.] Correlation of structural damage of R/C buildings with economic loss: calibration based on earthquakes of Athens 1999 and Thessaloniki 1978 [Conference] // 15th Concrete Conference. - Alexandroupoli : TGC-ETEK, 2006. Karabinis A.I. Assessment of seismic behaviour of r/c constructions Vulnerability and Risk [Conference] // 14th Concrete Conference. - Kos, Greece : [s.n.], 2003. Papaioanou Ch. and Voulgaris N et al The Utilization of New Seismological Data in the Compilation of the New Seismic Hazard Map of Greece [Conference] // 3th National Conference of Seismic Mechanics & Technical Seismology. - 2008. - GREEK. TCG - Staff Commitee Seismic retrofit existing buildings. Summary of results of 1st phase of research project in G.T.C. [Book]. - Athens : Technical Chamber of Greece, 2001. - GREEK. Vlachos I. Estimation of expected economic losses from earthquake to building wealth of the country and their consequences [Journal]. - [s.l.] : TCG - Technical Chamber of Greece, 1999. - GREEK. Websites Greek National Statistical Authority Buildings Inventory [Online] // GNSA. - http://www.statistics.gr. National Observatory of Athens [Online]. - 2001. - www.noa.gr. www.world-housing.net/ Newspaper articles Eleftherotypia The richters that changed us [Journal]. - Athens : Tegopoulos, 2009. - Daily Newspaper. TO VIMA Seismic vulnerable 8 out 10 buildings [Journal]. - Athens : [s.n.], 2009. Daily Newspaper Dimitris Detsis 10-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering -last page of main body- the appendices are attached- 10-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report PART V APPENDICES A. Model I-O volume and calculations B. Sensitivity analysis C. Choice of hazard curve D. Building behaviour assessment methods E. Building typologies Vulnerability curves F. Description of Greek building stock G. Retrofit Strategies H. Field research I. Personal Scope and Principles J. Societal & Individual Seismic Risk Estimator Model (s.i.s.r.e.m.) (excel file) Dimitris Detsis 10-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering -empty page- 10-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices A. Model input / output & calculations A.1 Main input output board (example for Athens) A.2 Main graphs for medium level decision maker A.2.1 Graph M-1 Societal risk of existing stock A.2.2 Graph M-2 Societal risk retrofit levels A.2.3 Graph M-3 Direct economic risk of existing stock A.2.4 Graph M-4 Direct economic retrofit stock A.3 Main graphs for top level decision maker A.3.1 Graph T-1 Societal risk of existing stock A.3.2 Graph T-2 Direct economic risk of existing stock A.3.3 Graph T-3 - individual risk & individual economic risk A.4 Calculation Volume Part I A.5 Calculation Volume Part II (upon requirement) Disclaimer Clause The best available sources and information has been used in preparing this report and these sources and information has been interpreted exercising all reasonable skill and care. Nevertheless, neither author nor reviewers accept any liability, whether direct, indirect or consequential, arising out of the provision of information in this report. Dimitris Detsis appendix A-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Societal & Individual Seismic Risk Estimator Model Brief User Guide The board on the right shows all the main parameters of the model. The user has only to enter municipality name. He can also change several parameters to have an overview of the sensitivity of parameters. The following table explains the contents of this board: name description source sheet INPUT DATA I 1 1 name municipality - user I-O 2 population - inventory S & I 3 area (km 2 ) total surface are of the municipality inventory S & I 4 buildings number of buildings in the municipality inventory S & I 5 building's area (m 2 ) total surface area of building - average 2 user I-O 6 occupancy rate (m 2 /prs) square meters that corresponds to each person calculated I 7 GDP per capita gross domestic product per person user I-O 8 interest rate interest rate for cost benefit analysis model to calculate MAX RETROFIT COST (5) user I-O 9 Societal risk limit lines on FN curve as criteria for risk acceptance, line changes depending to population 3 calculated L 10 Seismic zone hazard curve depending seismic zone according to the Greek Seismic Code (g) user I-O / H 11 H (factor) multiplies the hazard (for sensitivity analysis) user I-O 12 V (factor) multiplies the vulnerability (for sensitivity analysis) user I-O 13 rebuild cost / s.m. reference construction cost of dwellings per square meter ( ) user I-O 14 contents cost factor factor applied to (13) to determine contents loss user I-O 15 mortality coeff. coefficients which estimate mortality function user I-O 16 type of injuries type 0: to consider only deaths OR type 1: to add serious injuries as well user I-O 17 performance target degree of damage per exceedance frequency of the earthquake in every level user I-O / P INPUT DATA II 18 avg nr of floors average number of floors per building calculated I 19 coverage factor urban coverage factor of the municipality calculated I 20 urban character percentage of urban dwellings in the municipality inventory S 21 residential buildings percentage of residential buildings in the municipality inventory S 22 structural typologies graph with the distribution of the 7 typologies inventory I 23 structural vulnerability curves of exceedance probability of damage for the seven typologies calculated V 24 fatality rate for DS 5 indicative percent of fatalities when building collapses & casualties ratio graph calculated F 25 direct econ. Loss graph of direct economic loss calculated F OUTPUT DATA 1 AEF existing stock annual expected fatalities in the municipality calculated M 2 AEF exist/per capita annual expected fatalities per person in the municipality calculated M 3 AEL existing stock annual expected direct economic loss in the municipality calculated M 4 AEL exist/per capita annual expected direct economic loss per person in the municipality calculated M 5 individual risk (MAX) individual risk for the most exposed calculated M 6 MAX RETROFIT COST cost benefit analysis to estimate feasible maximum costs for each retrofit program calculated M GRAPHS FN CURVE representation of societal risk for existing stock calculated R retrofit influence averaged individual risk (output 1) / individual risk (MAX) (output 5) / annual expected calculated M losses per capita (output 4) for existing and 3 retrofit programs OTHER GRAPHS Graph M-1 Societal risk of existing stock FN Graph M-2 Societal risk retrofit levels FN Graph M-3 Direct economic risk of existing stock FL Graph M-4 Direct economic retrofit stock FL CALCULATION SHEETS 1 In red (bold): necessary parameters, while in black (regular): parameters that this model calculates, however the user has the choice to change them. 2 Enter the number and check if result (6) occupancy rate and (19) coverage factor have the right values. 3 Only indicatively here, please look at paragraph 9.6 of the final report appendix A-2 14 June 2010

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece OK INPUT / RESULTS BOARD OK Athina 11 code OK 1 OK 11 INPUT DATA II OUTPUT DATA Athina RETROFIT PROGRAMS 5987 18 avg nr of floors 33 3,3 1 AEF existing iti stock 43 4,3 Program 3 100% ALL MASONRY INPUT DATA I 15490 19 coverage factor 49% 2 AEF exist/per capita 5,E 06 Program 2 100% MS & RC85N / Y 1 name of municipality Athina 20 urban character 100% 3 AEL existing stock 420 mil Program 1 100% ALL (not RC95N) 2 population 789166 21 residential buildings 83% 4 AEL exist/per capita 532 3 area (km 2 ) 39 22 structural typologies 5 individual risk (MAX) 2,4E 05 OTHER STRATEGIES (TRIAL) 4 buildings 62277 6 MAX RETROFIT COST T=20yrs,NPV=0 deurbanization pop.: 100% 5 building's area (m 2 ) 1000 program 3 3523 mil Masonry 0% 100% 6 occupancy rate (m 2 /prs) 79 program 2 3719 mil R/C bef.85 0% 100% 7 GDP per capita 20000 program 1 4127 mil MS RCOLN RCOLY RC85N RC85Y RC95N RC95Y 8 interest rate 6,0% 9 Societal Risk Limit 23 structural vulnerability FN curve 1,E 01 1,E 02 1,E 03 I II III III 1,0E 05 3.200 1,E 04 40E04 4,0E 04 240 2,1E 03 80 1,E 05 1,4E 02 20 1 10 100 1.000 10.000 3,4E 02 10 10 seimsic zone in EAK2003 (a o (g)) I 0,16 11 H (factor) 1,0 12 V (factor) 1,0 13 rebuild cost / s.m. 800 14 contents cost factor 1/3 15 mortality coef. M3 0,65 M4 0,40 M5 0,70 16 type of casualties 0 17 performance targets level 2%/50yrs 36,2% 3 10%/50yrs 20,0% 0% 20% 40% 60% 50%/50yrs 7,0% 1,0E 01 1,0E 02 1,0E 03 1,0E 04 1,0E 05 casualties ratio 0% 10% 20% 30% 40% 50% 60% 24 fatalities rate for DS 5 3,01% 100,% 10,% 1,% 1% 0,1% 0,03% 0,01% 0,001% 1 for s.injuries 25 level 2%/50yrs 18,7% 2 10%/50yrs 10,0% 50%/50yrs 3,0% level 2%/50yrs 10,0% 1 10%/50yrs 5,0% 50%/50yrs 10% 1,0% repair cost /s.m. 1000 800 600 400 200 000 0,3% 0,003% 3,0% 30% 0% 20% 40% 60% 80% 100% direct economic loss 0% 20% 40% 60% 80% 100% damage degree * avg floor area (m 2 ) 303 ** pop.density (people per s.m) 0,203 Prepared by ddetsis 14/6/2010 Page 1 of 20 excceedance frequency averaged individual risk maximum individual risk 1,E 01 1,E 02 1,E 03 1,E 04 1,E 05 3,0E 05 2,5E 05 2,0E 05 1,5E 05 1,0E 05 5,0E 06 0,0E+00 1 10 100 1000 10000 N casualties retrofit influence 532 122 099 051 2,39E 05 3,64E 06 2,15E 06 1,49E 06 5,5E 06 2,1E 06 1,7E 06 11E 06 1,1E 06 existing stock retrofit program 3 retrofit program 2 retrofit program 1 600 500 400 300 200 100 000 annual expect loss / capita a ( )

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices empty page Dimitris Detsis appendix A-3

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece FN-CURVE 1 M-1 - societal risk of existing stock 1,E+00 1 MS 2 RCOLN 3 RCOLY 4 RC85N 5 RC85Y 1,E 01 annual exceedance probability (F ceedance probability (F bability (F 1/yr) yr) 6 RC95N 7 RC95Y 50%/20yrs ALL 50%/50yrs 1,E 02 10%/20yrs 10%/50yrs 1,E 03 1 E 03 2%/50yrs 1% / 100yrs 1% / 100yrs 1,E 04 1,E 05 1 10 100 1.000 10.000 N fatalities Prepared by ddetsis 14/6/2010 Page 2 of 20 Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece FN-CURVE 2 M-2 - societal risk - retrofit levels 1,E+00 existing stock I retrofit program 1 I retrofit program 2 retrofit program 2 I retrofit program 3 annual exceedance probability (F eedance probability (F ability (F 1/yrs) rs) 1,E 01 1,E 02 1 E 03 1,E 03 1,E 04 1,E 05 1 10 100 1.000 10.000 N fatalities Prepared by ddetsis 14/6/2010 Page 3 of 20

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece FL-CURVE 1 M-3 - direct economic risk - existing stock 1,0E 01 1 MS 2 RCOLN 20%/50yrs 3 RCOLY 4 RC85N 5 RC85Y annual exceedance probability (F ceedance probability (F bability (F 1/yr) yr) 1,0E 02 50%/50yrs 6 RC95N 7 RC95Y 7 RC95Y 10%/20yrs ALL 10%/50yrs 1,0E 03 2%/50yrs 1%/100yrs 1,0E 04 1,0E 05 0 2.000 4.000 6.000 8.000 10.000 12.000 14.000 16.000 18.000 20.000 direct economic losses - million Prepared by ddetsis 14/6/2010 Page 4 of 20 Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece FL-CURVE 2 M-4 - direct economic risk - retrofit stock 1,0E 01 existing retrofit program 1 20%/50yrs retrofit program 2 retrofit program 3 50%/50yrs annual exceedance probability (F ceedance probability (F bability (F 1/yr) yr) 1,0E 02 10%/20yrs 10%/50yrs 1,0E 03 2%/50yrs 1%/100yrs 1,0E 04 1,0E 05 0 2.000 4.000 6.000 8.000 10.000 12.000 14.000 16.000 18.000 20.000 direct economic losses million Prepared by ddetsis 14/6/2010 Page 5 of 20

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece TOP LEVEL 1 T-1 - societal risk of existing stock 1,E+00 1,E 01 Athina Thessaloniki Patra Pireas 50%/20yrs annual exceedance probability (F) 1,E 02 1,E 03 50%/50yrs 10%/20yrs 10%/50yrs 2%/50yrs 1,E 04 1E04 1% / 100yrs 1,E 05 1 10 100 1.000 10.000 N fatalities Prepared by ddetsis 14/6/2010 Page 6 of 20 TOP LEVEL 2 Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece T-2 - direct economic risk risk of existing stock 1,E 01 Athina Thessaloniki Patra Pireas 50%/20yrs 50%/50yrs annual exceedance probability (F) 1,E 02 1E 02 1,E 03 10%/20yrs 10%/50yrs 2%/50yrs 1,E 04 1% / 100yrs 1,E 05 0 2.000 4.000 6.000 8.000 10.000 12.000 14.000 16.000 18.000 20.000 L million Prepared by ddetsis 14/6/2010 Page 7 of 20

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece TOP LEVEL 3-4 expected loss s / capita 2000 1800 1600 1400 1200 1000 800 600 400 200 T 3 individual risk & individual economic risk 0 Pireas Patra Thessaloniki Athina annual expected loss/capita 1.005 1.746 420 532 avg. individual risk 6,E 06 7,E 06 4,E 06 5,E 06 max individual risk 2,50E 05 2,10E 05 2,20E 05 2,40E 05 3,E 05 3,E 05 2,E 05 2,E 05 1,E 05 5,E 06 0,E+00 probability of death of death Prepared by ddetsis 14/6/2010 Page 8 of 20 Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece CALCULATION VOLUME PART I Prepared by ddetsis 14/6/2010 Page 9 of 20

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece INVENTORY area Athina municipality urban parameters character main function seismic zone population p 789.166 7,22% total area 39 s.km. coverage factor 49% urban 62.277 100% residential 51.584 83% buildings 62.277 1,56% footprint 19 s.km. occupancy 79 m 2 /hab rural 0 0% non-resid 10.693 17% reference pga - a 0 0,16 floors avg. floor material construction date structural typologies number % GDP ground 11.075 18% number 205227 ms 10.227 16% bef '85 56.495 91% 1 MS 10.227 16% per capita 20000 plus 1 13.005 21% per build. 3,3 rc 51.913 83% 85-'95 3.522 6% 2 RCOLN 43.534 70% IRR 0,06 plus 2 7.971 13% area 303 s.m. other 137 0% after '95 2.260 4% 3 RCOLY 2.613 4% 3_5 23.660 38% 46.268 89% 4 RC85N 3.314 5% LIMIT over 6 6.566 11% 3.522 7% 5 RC85Y 199 0% F N pilotis 2.939 6% 2.260 4% 6 RC95N 2.263 4% 1,0E-05 3.200 7 RC95Y 128 0% 4,0E-04 04 240 25.000 2,1E-03 80 floors 60.000 material 1,4E-02 20 20.000 3,4E-02 10 50.000 4,E-01 1,85 15.000 40.000 10.000 5.000 0 ground plus 1 plus 2 3_5 over 6 30.000 20.000 10.000 0 ms rc other 60.000 50.000 construction date 50.000 40.000 structural typologiestypologies 40.000 30.000 20.000 10.000 30.000 20.000000 10.000 0 bef '85 85 '95 after '95 0 MS RCOLN RCOLY RC85N RC85Y RC95N RC95Y Prepared by ddetsis 14/6/2010 Page 10 of 20 Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece SEISMIC HAZARD P ( A g a g ) P R,1 F [Papazachos et al,1993] annual time return exceedance peak groun acc. (a g ) 1 span period frequency log 10 Y=C 1 *log 10 T R +C 2 P R T L T R P R,1 ZONE-I ZONE-II ZONE-III 1% 10000 994992 1,0E-06 0,93 1,24 1,66 1% 1000 99499 1,0E-05 0,55 0,76 1,05 1% 100 9950 1,0E-04 0,32 0,46 0,66 2% 100 4950 2,0E-04 04 0,28 0,40 0,57 2% 50 2475 #### 0,23 0,34 0,50 10% 100 949 1,1E-03 0,19 0,28 0,41 10% 50 475 2,1E-03 0,16 0,24 0,36 20% 50 224 4,5E-03 0,14 0,20 0,31 10% 20 190 5,3E-03 0,13 0,20 0,30 20% 20 90 1,1E-02 0,11 0,17 0,26 50% 50 72 1,4E-02 0,10 0,16 0,25 2% 1 49 2,0E-02 0,10 0,15 0,23 C1 C2 50% 20 29 3,4E-02 0,08 0,13 0,21 ZONE I 0,230 1,590 10% 1 9 1,0E-01 0,0707 0,10 0,16 ZONE II 0,215 1,805 50% 1 1 5,0E-01 0,04 0,07 0,11 ZONE III 0,200 2,020 1,0E 01 annual exceedance probability of a g annual excedance probability - P r,1 50%/50yrs 1,0E 02 10%/50yrs 1,0E 03 2%/50yrs 1,0E 04 ZONE I ZONE III 1,0E 05 2%/50yrs 10%/50yrs 50%/50yrs 0,00 000 008 0,08 016 0,16 024 0,24 032 0,32 040 0,40 048 0,48 056 0,56 064 0,64 072 0,72 080 0,80 088 0,88 096 0,96 104 1,04 0 4,0E-04 2,1E-03 1,4E-02 peak ground accelaration - a g (g) Prepared by ddetsis 14/6/2010 1,04 4,0E-04 2,1E-03 Page 1,4E-02 11 of 20 ZONE II

SEISMIC VULNERABILITY OF EXISTING BUILDINGS d f( a ) [EPANTYK - TCG, 2007] Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece i g 1 1 1 2 1 3 1 4 1 5 16 17 1 a g g/ /a o MS MS-q RCOLN RCOLN-q RCOLY RCOLY-q RC85N RC85N-q RC85Y RC85Y-q RC95N RC95N-q RC95Y RC95Y-q 0,00 0 0% 0 0% 0 0% 0 0% 0 0% 0 0% 0 0% 0,50 10% 9% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 1,00 15% 19% 8% 5% 8% 6% 4% 2% 6% 3% 2% 1% 4% 3% 1,50 30% 29% 10% 11% 12% 13% 5% 6% 8% 8% 4% 3% 8% 8% 2,00 40% 38% 15% 17% 16% 20% 9% 11% 10% 13% 6% 5% 10% 14% quadratic: d=c 1 *(a g /a 0 ) 2 +c 2 *(a g /a 0 )+c 3 2,50 47% 46% 20% 22% 25% 27% 10% 16% 16% 19% 8% 7% 16% 20% 3,00 55% 54% 30% 28% 35% 34% 25% 21% 25% 25% 10% 9% 30% 26% 5,00 80% 79% 55% 53% 70% 66% 50% 46% 60% 56% 18% 17% 60% 58% 7,00 100% 95% 80% 81% 100% 102% 75% 77% 95% 97% 26% 25% 95% 96% c 11 1,1 c 21 2,1 c 12 1,2 c 22 2,2 c 13 1,3 c 23 2,3 c 14 1,4 c 24 2,4 c 15 1,5 c 25 2,5 c 16 1,6 c 26 2,6 c 17 1,7 c 27 2,7-1,22 22,37 0,28 10,29 0,42 12,64 0,71 6,74 1,20 5,89 0,00 4,00 0,97 7,81 c 3,1-2,23 c 3,2-5,08 c 3,3-7,33 c 3,4-5,30 c 3,5-3,62 c 3,6-2,60 c 3,7-5,77 100% vulnerability curves 80% damage age degree - d i 60% 40% 20% 0% 0,0 1,0 2,0 3,0 4,0 5,0 6,0 7,0 a g /a 0 MS q RCOLN q RCOLY q RC85N q RC85Y q RC95N q RC95Y q Prepared by ddetsis 14/6/2010 Page 12 of 20 EXCEEDANCE PROBABILITY OF DAMAGE FOR TYPOLOGIES Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece a o = 0,16 P R,1 PGA a g g/ /a o 1-MS 2-RCOLN 3-RCOLY 4-RC85N 5-RC85Y 6-RC95N 7-RC95N 1,0E-06 0,93 5,83 87% 64% 80% 80% 72% 21% 73% 1,0E-05 0,55 3,43 60% 34% 41% 26% 31% 11% 32% 1,0E-04 0,32 2,02 38% 17% 20% 11% 13% 5% 14% 2,0E-04 0,28 1,72 33% 13% 16% 8% 10% 4% 11% 4,0E-04 04 0,23 1,47 28% 11% 12% 6% 8% 3% 8% 1,1E-03 0,19 1,18 22% 7% 8% 4% 5% 2% 5% 2,1E-03 0,16 1,00 19% 6% 6% 2% 3% 1% 3% 4,5E-03 0,14 0,84 16% 4% 4% 1% 2% 1% 2% 5,3E-03 0,13 0,81 15% 3% 3% 1% 2% 1% 1% 11E-02 1,1E-02 0,11 0,68 12% 2% 2% 0% 1% 0% 0% 1,4E-02 0,10 0,65 12% 2% 1% 0% 1% 0% 0% 2,0E-02 0,10 0,60 11% 1% 0% 0% 0% 0% 0% 3,4E-02 0,08 0,53 9% 0% 0% 0% 0% 0% 0% 1,0E-01 0,07 0,41 7% 0% 0% 0% 0% 0% 0% 50E-01 5,0E-01 004 0,04 026 0,26 4% 0% 0% 0% 0% 0% 0% 1,0E+00 annual exceedance probability of damage d i annual al excedance probability - P r,1 1,0E 01 1,0E 02 1,0E 03 10E04 1,0E 04 50%/50yrs 10%/50yrs 2%/50yrs 1 MS 2 RCOLN 3 RCOLY 4 RC85N 5 RC85Y 6 RC95N 7 RC95N 1,0E 05 2%/50yrs 10%/50yrs 50%/50yrs 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% damage degree - % 0% 4,0E-04 2,1E-03 1,4E-02 100% 4,0E-04 2,1E-03 1,4E-02 Prepared by ddetsis 14/6/2010 Page 13 of 20

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece DIRECT ECONOMIC LOSS FUNCTION L i f ( d ) rebuild costs 800 repair cost ~ damage [Kappos, A.I. et al, 2006] 1,65-1,70 1,20 0,01 1% MS 800 per m 2 L buil. = (0,8725*d 3-0,2178*d 2 +0,3806*d+0,1385)xRCi, 0,8725-0,2178 0,3806 0,1385 R/C 800 per m 2 direct economic loss function contents contents loss ~ damage 1000 0,33 264 per m 2 L contents = d*contents 800 DAMAGE DEGREE repair costs contents 600 state description range mean L buil. MS L buil. RC L contents DS-1 slight 0% 1% 0,5% 000 000 000 400 0,5% 112 DS-2 moderate 1% 10% 6% 057 057 015 6% 127 200 [Kappos et al] DS-3 heavy 10% 30% 20% 156 156 053 20% 170 DS-4 v. heavy 30% 60% 45% 285 285 119 000 45% 276 DS-5 collapse 60% 100% 80% 581 581 211 0% 20% 40% 60% 80% 80% 100% 600 *monetary values in per s.m. of built area 100% 928 928 264 damage degree 100% 939 repair cost /s.m. MORTALITY FUNCTION i f ( d ) [Coburn & Spence, 2002]: Ks=F*[M1*M2*M3*(M4+M5*(1-M4))] casualties function Seriously injured - Trapped in collapsed area 100,% M 2 occupancy distribution in day M 3 ratio of occupants trapped 30% 0,57 rural resid non-resid 0,65 for r/c buildings (3-5 floors) 10,% 45% 60% 40% 3% Fatalities 0% 0,0003% M4 deaths at t=0 M 5 deaths at t=36h 1,% 1% 0,0003% 0,3% 040 0,40 070 0,70 1% 0,003% 003% 0,1% 10% 0,003% DAMAGE DEGREE F collapse CASUALTIES 0,03% 10% 0,03% state description range mean area* S.Injuries Fatalities 0 30% 0,03% 0,01% DS-1 slight 0% 1% 1% 0,001% 0,0004% 0,0003% 0,0003% 30% 0,3% 0,003% DS-2 moderate 1% 10% 6% 0,01% 01% 0,004% 0,003% 0,003% 60% 0,3% DS-3 heavy 10% 30% 20% 0,1% 0,04% 0,03% 0,03% 0,001% 60% 3,0% DS-4 v. heavy 30% 60% 45% 1% 0,4% 0,3% 0,3% 0% 20% 40% 60% 100% 80% 100% 3% DS-5 collapse 60% 100% 80% 10% 4% 3% 3% damage degree 100% 30% 100% 100% 37% 30% 30% casualties alties rate Prepared by ddetsis 14/6/2010 Page 14 of 20 PERFORMANCE TARGETS FOR RETROFITTED BUILDINGS ESTIMATION OF SOCIETAL RISK AND ECONOMIC LOSS FOR SEISMIC RISK IN ATHENS time return annual REFERENCE PGA RETROFIT LEVELS - span period frequency PERFORMANCE TARGETS P R T L T R P R,1 ZONE-I ZONE-II ZONE-III LEVEL 1 LEVEL 2 LEVEL 3 1% 10000 994992 1,0E-06 0,93 1,24 1,66 21% 39% 73% 1% 1000 99499 1,0E-05 0,55 0,76 1,05 16% 30% 57% 1% 100 9950 1,0E-04 0,32 0,46 0,66 11% 21% 41% 2% 100 4950 2,0E-04 04 0,28 0,40 0,57 10% 19% 36% 2% 50 2475 4,0E-04 0,23 0,34 0,50 9% 16% 31% 10% 100 949 1,1E-03 0,19 0,28 0,41 6% 13% 25% 10% 50 475 2,1E-03 0,16 0,24 0,36 5% 10% 20% 20% 50 224 4,5E-03 0,14 0,20 0,31 3% 7% 15% 10% 20 190 5,3E-03 0,13 0,20 0,30 3% 7% 14% 20% 20 90 1,1E-02 0,11 0,17 0,26 1% 4% 8% 50% 50 72 1,4E-02 0,10 0,16 0,25 1% 3% 7% excedance probability - P r,1 PERFORMANCE TARGETS OF RETROFIT 50%/50yrs 10%/50yrs damage degree - % Prepared by ddetsis 14/6/2010 Page 15 of 20 excedance edance probability ity - P r,1 1,E 01 1,E 02 1,E 03 1E04 1,E 04 2%/50yrs 2% 1 49 2,0E-02 0,10 0,15 0,23 0% 2% 4% C1 C2 50% 20 29 3,4E-02 0,08 0,13 0,21 0% 0% 1% 1,E 05 2%/50yrs 10%/50yrs 50%/50yrs LEVEL I -0,081-0,021 10% 1 9 1,0E-01 0,0707 0,10 0,16 0% 0% 0% 0% 20% 40% 40E04 21E03 14E02 0% 4,0E-04 2,1E-03 1,4E-02 LEVEL II 60% -0,130-0,037 80% 037 damage degree - % 50% 1 1 5,0E-01 0,04 0,07 0,11 0% 0% 0% 100% 4,0E-04 2,1E-03 1,4E-02 LEVEL III -0,227-0,069 1,0E 01 1,0E 02 1,0E 03 1,0E 04 1,0E 05 COMPARISON OF DAMAGE EXCEEDANCE PROBABILITY FOR EXISTING & RETROFITTED BUILDINGS LEVEL 2 LEVEL 1 LEVEL 3 1 MS 2 RCOLN 3 RCOLY 4 RC85N 5 RC85Y 6 RC95N 7 RC95N 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

ESTIMATION OF SOCIETAL RISK AND ECONOMIC LOSS FOR SEISMIC RISK IN ATHENS RETROFITTED "PERFORMANCE" - (VULNERABILITY OF RETROFITTED TYPOLOGIES) - 1 exceedance probability of damage degree of retrofitted typologies level 3 P R,1 1-MS-R3 2-RCOLN-R3 3-RCOLY-R3 4-RC85N-R3 5-RC85Y-R3 6-RC95N-R3 7-RC95Y-R3 1,0E-06 73% 64% 73% 73% 72% 21% 73% 1,0E-05 57% 34% 41% 26% 31% 11% 32% 1,0E-04 38% 17% 20% 11% 13% 5% 14% 2,0E-04 04 33% 13% 16% 8% 10% 4% 11% 4,0E-04 28% 11% 12% 6% 8% 3% 8% 1,1E-03 22% 7% 8% 4% 5% 2% 5% 2,1E-03 19% 6% 6% 2% 3% 1% 3% 4,5E-03 15% 4% 4% 1% 2% 1% 2% 53E-03 5,3E-03 14% 3% 3% 1% 2% 1% 1% 1,1E-02 8% 2% 2% 0% 1% 0% 0% 1,4E-02 7% 2% 1% 0% 1% 0% 0% 2,0E-02 4% 1% 0% 0% 0% 0% 0% 3,4E-02 1% 0% 0% 0% 0% 0% 0% 1,0E-01 0% 0% 0% 0% 0% 0% 0% 5,0E-01 0% 0% 0% 0% 0% 0% 0% level 2 P R,1 1-MS-R2 2-RCOLN-R2 3-RCOLY-R2 4-RC85N-R2 5-RC85Y-R2 6-RC95N-R2 7-RC95Y-R2 1,0E-06 39% 39% 39% 39% 39% 21% 39% 10E05 1,0E-05 30% 30% 30% 26% 30% 11% 30% 1,0E-04 21% 17% 20% 11% 13% 5% 14% 2,0E-04 19% 13% 16% 8% 10% 4% 11% 4,0E-04 16% 11% 12% 6% 8% 3% 8% 1,1E-03 13% 7% 8% 4% 5% 2% 5% 21E03 2,1E-03 10% 6% 6% 2% 3% 1% 3% 4,5E-03 7% 4% 4% 1% 2% 1% 2% 5,3E-03 7% 3% 3% 1% 2% 1% 1% 1,1E-02 4% 2% 2% 0% 1% 0% 0% 1,4E-02 3% 2% 1% 0% 1% 0% 0% 20E02 2,0E-02 2% 1% 0% 0% 0% 0% 0% 3,4E-02 0% 0% 0% 0% 0% 0% 0% 1,0E-01 0% 0% 0% 0% 0% 0% 0% 5,0E-01 0% 0% 0% 0% 0% 0% 0% exceedance probability exceedance probability 1,0E 01 1,0E 02 1,0E 03 10E04 1,0E 04 1,0E 05 exceedance probability of damage di retrofit level 3 1 MS R3 2 RCOLN R3 3 RCOLY R3 4 RC85N R3 5 RC85Y R3 6 RC95N R3 7 RC95Y R3 0% 20% 40% 60% 80% damage degree d ri exceedance probability of damage di 1,0E 01 retrofit level 2 1 MS R2 2 RCOLN R2 1,0E 02 3 RCOLY R2 1,0E 03 1,0E 04 4 RC85N R2 5 RC85Y R2 6 RC95N R2 7 RC95Y R2 1,0E 05 0% 20% 40% 60% 80% damage degree d ri Prepared by ddetsis 14/6/2010 Page 16 of 20 ESTIMATION OF SOCIETAL RISK AND ECONOMIC LOSS FOR SEISMIC RISK IN ATHENS RETROFITTED "PERFORMANCE" - (VULNERABILITY OF RETROFITTED TYPOLOGIES) - 2 exceedance probability of damage degree of retrofited typologies level 1 exceedance probability of damage di retrofit level 1 P R,1 1-MS-R1 2-RCOLN-R1 3-RCOLY-R1 4-RC85N-R1 5-RC85Y-R1 6-RC95N-R1 7-RC95Y-R1 1,0E-06 21% 21% 21% 21% 21% 21% 21% 1,0E-05 16% 16% 16% 16% 16% 11% 16% 1,0E-04 11% 11% 11% 11% 11% 5% 11% 2,0E-04 04 10% 10% 10% 8% 10% 4% 10% 4,0E-04 9% 9% 9% 6% 8% 3% 8% 1,1E-03 6% 6% 6% 4% 5% 2% 5% 2,1E-03 5% 5% 5% 2% 3% 1% 3% 4,5E-03 3% 3% 3% 1% 2% 1% 2% 53E-03 5,3E-03 3% 3% 3% 1% 2% 1% 1% 1,1E-02 1% 1% 1% 0% 1% 0% 0% 1,4E-02 1% 1% 1% 0% 1% 0% 0% 2,0E-02 0% 0% 0% 0% 0% 0% 0% 3,4E-02 0% 0% 0% 0% 0% 0% 0% 1,0E-01 0% 0% 0% 0% 0% 0% 0% 5,0E-01 0% 0% 0% 0% 0% 0% 0% exceedance eedance probability 1,0E 01 1 MS R1 2 RCOLN R1 1,0E 02 3 RCOLY R1 4 RC85N R1 1,0E 03 5 RC85Y R1 6 RC95N R1 1,0E 04 7 RC95Y R1 1,0E 05 0% 20% 40% 60% 80% damage degree d ri Prepared by ddetsis 14/6/2010 Page 17 of 20

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece SOCIETAL & ECONOMICAL RISK FOR EXISTING & RETROFITTED STOCK ALL LOSS avg floor floor area build. area rebuild cost / contents inventory ALL FATALITIES occupancy rate Athina municipality 3,3 303 s.m. 1000 s.m. 800 -s.m. 264 -s.m. 62.277 2% Athina municipality 0,0127 hab/m 2 exc. probability cost casualties cost casualties cost casualties cost casualties P r,1 L all area N all area L all area N all area million N million N million N million N 1,0E-06 40.164 22.934 37.293 22.934 21.672 2.297 13.680 234 1,0E-05 21.455 5.700 21.136 2.183 17.471 234 11.075 234 1,0E-04 04 12.674 577 12.674 577 11.355 226 8.248 234 2,0E-04 10.695 566 10.695 566 9.567 215 7.298 20 4,0E-04 8.952 215 8.952 215 7.960 215 6.340 20 1,1E-03 6.859 56 6.859 56 5.982 56 4.992 20 2,1E-03 5.549 56 5.549 56 4.707 20 3.996 20 4,5E-03 4.200 55 4.110 55 3.352 19 2.833 19 5,3E-03 3.960 55 3.822 55 3.105 19 2.596 19 1,1E-02 2.958 55 2.555 19 2.041 19 1.500 19 1,4E-02 2.695 55 2.201 19 1.752 19 0 1 2,0E-02 2.233 55 1.564 19 1.235 19 0 1 3,4E-02 1.128 4 0 1 0 1 0 1 1,0E-01 862 4 0 1 0 1 0 1 5,0E-01 516 4 0 1 0 1 0 1 LEVEL 3 LEVEL 2 L all area N all area LEVEL 1 L all area N all area 1,0E+00 FL curve 1,E+00 FN curve 1,0E 01 1,E 01 1,0E 02 1,E 02 frequency 1,0E 03 frequency 1,E 03 1,0E 04 1,E 04 1,0E 05 1,E 05 1,0E 06 0 10.000 20.000 30.000 40.000 50.000 DIRECT ECONOMIC LOSS milllion 1,E 06 1 10 100 1000 10000 100000 N casualties Prepared by ddetsis 14/6/2010 Page 18 of 20 ANNUAL EXPECTED VALUES - SOCIETAL & ECONOMICAL RISK Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece ALL E.LOSS avg floor floor area build. area rebuild cost / contents inventory ALL FATALITIES occupancy rate Athina municipality 3,3 303 s.m. 1000 s.m. 800 -s.m. 264 -s.m. 62.277 2% Athina municipality 0,0127 hab/m 2 exc. probability probability cost casualties cost casualties cost casualties cost casualties P r,1 P area L all area N all area L all area N all area L all area N all area L all area N all million N million N million N million N 1,0E-06 1,0E-06 40.164 22.934 37.293 22.934 21.672 2.297 13.680 234 1,0E-05 9,0E-06 21.455 5.700 21.136 2.183 17.471 234 11.075 234 1,0E-04 9,1E-05 12.674 577 12.674 577 11.355 226 8.248 234 2,0E-04 1,1E-04 10.695 566 10.695 566 9.567 215 7.298 20 4,0E-04 2,9E-04 8.952 215 8.952 215 7.960 215 6.340 20 1,1E-03 7,6E-04 6.859 56 6.859 56 5.982 56 4.992 20 2,1E-03 1,3E-03 5.549 56 5.549 56 4.707 20 3.996 20 4,5E-03 3,1E-03 4.200 55 4.110 55 3.352 19 2.833 19 5,3E-03 2,1E-03 3.960 55 3.822 55 3.105 19 2.596 19 1,1E-02 8,9E-03 2.958 55 2.555 19 2.041 19 1.500 19 1,4E-02 4,8E-03 2.695 55 2.201 19 1.752 19 0 1 2,0E-02 1,5E-02 2.233 55 1.564 19 1.235 19 0 1 3,4E-02 1,9E-02 1.128 4 0 1 0 1 0 1 1,0E-01 8,1E-02 862 4 0 1 0 1 0 1 5,0E-01 4,2E-01 516 4 0 1 0 1 0 1 annual expected AEL exist AEF exist AEL 3 AEF 3 AEL 2 AEF 2 AEL 1 AEF1 loss fatalities 420 4,3 96 1,7 78 1,3 40 0,9 annual expected AEL exist/capita AEF exist/capita AEL 3/capita AEF 3/capita AEL 2/capita AEF 2/capita AEL 1/capita AEF 1/capita per capita (in ) 532 55E-06 5,5E-06 122 21E-06 2,1E-06 099 17E-06 1,7E-06 051 11E-06 1,1E-06 % of GDP/capita 3% 1% 0% 0% individual risk (max) 2,39E-05 3,64E-06 2,15E-06 1,49E-06 STOCK EXISTING LEVEL L 3 LEVEL L 2 LEVEL L 1 COST BENEFIT ANALYSIS (maximum retrofit cost) Investment horizon t= For NPV= 0 IRR= 6,0% 20 yrs mil. level 3 level 2 level 1 1st year 306 323 358 Total Inv. 3523 3719 4127 millions 450 400 350 300 250 200 150 100 50 0 annual expected values (AEF & AEL) 420 4,30 1,67 96 1,35 78 0,88 40 existing stock retrofit program 3 retrofit program 2 retrofit program 1 5,00 4,00 3,00 2,00 200 1,00 0,00 fatalities Prepared by ddetsis 14/6/2010 Page 19 of 20

PROPOSAL FOR SOCIETAL RISK LIMITS Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece [Vrijling et al.] probability/timespan/return period societal risk limit =1 =0,1 -I 1% 100 9950 1,0E 04 1 I II III 1 2,8E+01 2,8E-01 1,E 01 10E04 1,0E 04 25000 10 2,8E-01 2,8E-03 2% Vrijling et al. =1 50 2475 4,0E 04 1 100 2,8E-03 2,8E-05 4,0E 04 25000 1000 2,8E-05 2,8E-07 10% 50 475 2,1E 03 1 10000 2,8E-07 2,8E-09 2,1E 03 25000 10% 20 190 5,3E 03 1 =0,1 combination 1,E 02 5,3E 03 25000 I I I 50% 50 72 1,4E 02 1 1,0E-05 180 1600 3200 Vrijling et al. =0,1 1,4E 02 25000 4,0E-04 30 120 240 50% 20 29 3,4E 02 1 2,1E-03 13 40 80 3,4E 02 25000 14E02 1,4E-02 5 10 20 20% 1 4 2,0E 01 1 3,4E-02 3,25 5 10 1,E 03 2,0E 01 25000 4,0E-01 1 1 1,85 annual exceedance probability (F) proposal population limit 10E-05 1,0E-05 40E-04 4,0E-04 21E-03 2,1E-03 14E-02 1,4E-02 34E-02 3,4E-02 10000 50000 I I 180 30 13 5 3,25 1 50000 500000 I II 1,E 04 1600 120 40 10 5 1 500000 5000000 III III 3200 240 80 20 10 1,85 50% / 20yrs 50% / 50yrs 10% / 50 yrs 2% / 50 yrs 1% / 100 yrs 1,E 05 1 10 100 1.000 10.000 N fatalities Prepared by ddetsis 14/6/2010 Page 20 of 20

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices B. Sensitivity analysis This appendix presents comparison graphs from sensitivity analysis (chapter 8) Dimitris Detsis appendix B-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering -empty page- appendix B-2 14 June 2010

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece SENSITIVITY ANALYSIS 1 1,E+00 1.1 influence of hazard on FN curve 1,E 01 1.2 influence of hazard on FL curve 1,E 01 frequency F 1,E 02 1,E 03 300% 140% 120% 80% 60% frequency F 1,E 02 1,E 03 300% 140% 120% 80% 60% 1,E 04 30% 100% 1,E 04 30% 100% 1,E 05 1 10 100 1.000 10.000 fatalities N 1,E 05 0 10.000 20.000 30.000 direct economic loss L (million ) 1,E+00 1.3 influence of vulnerability on FN curve 1,E 01 1.4 influence of vulnurability on FL curve 1,E 01 1E01 300% 1,E 02 300% frequency ency F 1,E 02 1,E 03 140% 120% 80% 60% frequency ency F 1,E 03 140% 120% 80% 60% 1,E 04 30% 100% 1,E 04 30% 100% 1,E 05 1 10 100 1.000 10.000 fatalities N 1,E 05 0 10.000 20.000 30.000 direct economic loss L(million ) Prepared by ddetsis 14/6/2010 Page 1 of 7 SENSITIVITY ANALYSIS 2 Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece 1,E+00 2.1 influence of M3 on FN curve 1,E+00 2.3 influence of M5 on FN curve 1,E 01 1,E 01 frequency F 1,E 02 1,E 03 0,91 0,78 0,52 0,39 frequency F 1,E 02 1,E 03 0,98 0,84 0,56 0,42 1,E 04 1E04 0,33 065 0,65 1,E 04 1E 04 0,21 070 0,70 1,E 05 1 10 100 1.000 10.000 fatalities N 1,E 05 1 10 100 1.000 10.000 fatalities N frequency ency F 1,E+00 1,E 01 1,E 02 1,E 03 1,E 04 2.2 influence of M4 on FN curve 0,80 0,56 0,48 0,32 0,24 0,33 040 0,40 annual expected fatalities 6 5 4 3 2 1 2.4 influence of M3,M4 & M5 on AEF occupancy distribution in day (M3) in day deaths at t=0 (M4) deaths at t=36h (M5) 1,E 05 1 10 100 1.000 10.000 fatalities N 0 0,0 0,2 0,4 0,6 0,8 1,0 rate value Prepared by ddetsis 14/6/2010 Page 2 of 7

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece SENSITIVITY ANALYSIS 3 1,E 01 3.1 influence of avg building area on FL curve 1,E+00 3.2 influence of population on FN curve 1,E 01 frequency F 1,E 02 1,E 03 1,E 04 140% 120% 80% 60% 30% 100% frequency F 1,E 02 1,E 03 1,E 04 120% 110% 90% 80% 100% 1,E 05 0 5.000 10.000 15.000 20.000 25.000 30.000 direct economic loss L (million ) 1,E 05 1 10 100 1.000 10.000 fatalities N AEL 3.3 AEL occupancy rate avg building area 120 400 110 350 100 300 90 250 80 200 AEL 150 occupancy rate 70 60 50 100 40 700 800 900 1.000 1.100 1.200 1.300 1.400 1.500 avg building area (s.m.) urban coverage ranges 35% 75% occupancy rate people / s.m. ate people / s.m. Prepared by ddetsis 14/6/2010 Page 3 of 7 SENSITIVITY ANALYSIS 4 Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece 1,E+00 4.1 influence of NR of buildings on FN curve 1,E 01 4.2 influence of NR of buildings on FL curve 1,E 01 1,E 02 frequency F 1,E 02 1,E 03 80% 60% 20% frequency F 1,E 03 80% 60% 20% 1,E 04 100% 1E 04 1,E 04 100% 1,E 05 1 10 100 1.000 10.000 fatalities N 1,E 05 0 5.000 10.000 15.000 20.000 direct economic loss i L (million ) 1,E+00 4.3 influence of deurbanism on FN curve 1,E 01 4.4 influence deurbanism on FL curve 1,E 01 1,E 02 frequency ency F 1,E 02 1,E 03 20% 80% 60% 40% frequency ency F 1,E 03 80% 60% 40% 20% 1,E 04 100% 1,E 04 100% 1,E 05 1 10 100 1.000 10.000 fatalities N 1,E 05 0 5.000 10.000 15.000 20.000 direct economic loss L(million ) Prepared by ddetsis 14/6/2010 Page 4 of 7

CONCLUSIONS - SOCIETAL RISK ESTIMATION Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece annual expected fatalities 20 15 10 5 5.1 sensitivity of AEF hazard vulnerability occupancy distribution in day (M3) deaths at t=0 (M4) deaths at t=36h (M5) nr of buildings de urbanism population annual expected fatalities 5,0 4,5 4,0 3,5 3,0 2,5 5.2 sensitivity of AEF zoomed hazard vulnerability occupancy distribution in day (M3) deaths at t=0 (M4) deaths at t=36h (M5) nr of buildings de urbanism 0 0% 50% 100% 150% 200% parameter change 2,0 population 70% 80% 90% 100% 110% 120% 130% parameter change 1,E+00 1,E 01 5.3 sensitivity on FN curve hazard vulnerability occupancy distribution in day (M3) population frequency ency F 1,E 02 1,E 03 nr of buildings de urbanism 100% 1,E 04 1,E 05 1 10 100 1.000 10.000 fatalities N Prepared by ddetsis 14/6/2010 Page 5 of 7 CONCLUSIONS - DIRECT ECONOMIC RISK ESTIMATION Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece annual expected loss million 1000 800 600 400 200 6.1 sensitivity of AEL hazard 300 vulnerability cost per s.m. avg building area nr of buildings de urbanism annual expected loss al expected loss s million 250 200 150 6.2 sensitivity of AEL hazard vulnerability cost per s.m. avg building area nr of buildings de urbanism 0 0% 50% 100% 150% 200% parameter change 100 70% 80% 90% 100% 110% 120% 130% parameter change 1,E 01 6.3 influence deurbanism on FL curve frequency F 1,E 02 1,E 03 1E 03 1,E 04 hazard vulnerability cost per s.m. avg building area nr of buildings de urbanism 100% 1,E 05 0 5.000 10.000 15.000 20.000 direct economic loss L(million ) Prepared by ddetsis 14/6/2010 Page 6 of 7

Estimation of Societal Risk and Direct Economic Loss from Earthquakes in Greece CONCLUSIONS - RETROFIT LEVELS 120 Retrofit levels 2,0 annual expect loss in millions loss in millions 100 80 60 40 20 1,5 1,0 0,5 annual expected fatalities 0 30% 10% 20% 7% 15% 5% 10% 3% 5% 1% retrofit level 0,0 Prepared by ddetsis 14/6/2010 Page 7 of 7

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices C. Choice of hazard curve This appendix presents graphs which guide to the choice of hazard curve (chapter 7) Figure C-1 - PSHA for Athens (Tsapanos, 2004) Figure C-2 comparison between hazard of equation 7-1 and (Tsapanos, 2004) for Athens Dimitris Detsis appendix C-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Figure C-3 influence of different PSHA in FN-curve Figure C-4 influence of different PSHA in FL-curve Papaioanou Ch. and Voulgaris N et al The Utilization of New Seismological Data in the Compilation of the New Seismic Hazard Map of Greece [Conference] // 3th National Conference of Seismic Mechanics & Technical Seismology. - 2008. - GREEK. Tsapanos T.M., Mäntyniemi, P. & Kijko, A A probabilistic seismic hazard assessment for Greece and the surrounding region including site-specific considerations [Journal] // Annals of geophysics. - December 2004. - 6 : Vol. 47. - pp. 1675-1688. appendix C-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices D. Building behaviour assessment methods This appendix analyses reasons to asses buildings and methodologies. It deepens in analytical methods, pushover analysis, and performance-based design principles. (related to chapter 3 and chapter 5) D.1 Reasons to asses There are two reasons for existing building assessment and retrofit. The first is deterioration of their strength due to ageing and the second synchronization with recent safety standards. Especially in the case of buildings subjected to earthquakes, more reasons increase the necessity for assessment of their structural safety. These are: identification of the pathodology of the building it is not simple, in order to estimate structural safety you will need to model the whole structural system technological developments are faster both for action and strength estimation, for example in Greece 32% are built before 1960 where no seismic code was enacted construction practices have been proved to increase seismic vulnerability, for example pilotis (soft ground floor) Thus, assessment of the structural behaviour is important in order to decide between demolish/retrofit/do nothing. D.2 Methods There are 3 clusters of method to assess structural behaviour against earthquakes: method description use empirical classification of vulnerability depending on basic characteristics such age, building code, materials, structural system, ground large sets of buildings seismic risk policy multi-criteria evaluation scoring criteria with certain weights, those criteria are more detailed than previous and focus on structural members, which are important for building behaviour small sets of buildings preliminary check analytical detailed structural analysis using modelling individual building mapping, analysis, retrofit design Table D-1 assessment methodologies (Karabinis, 2003) For empirical methods, seismic action is determined in terms of micro-seismic intensity or peak ground acceleration. However, for analytical methods, except elastic response spectrum (see par. 5.2.3), and for more precise analysis real or engineer-composed acceleration-graphs (time-history) can be applied. Since empirical methodology / vulnerability estimation is described in Chapter 5, analytical methods are presented here in more detail. Dimitris Detsis appendix D-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering However, before some details are given about multi-criteria evaluation methodologies. In those, engineer is tracing on site the condition of critical members of the structure. These are assessed qualitative depending on their capacity to bear seismic load. Then a scoring board is completed and applying weights on the different criteria, a final score is derived. The final score can lead to decision that analytical assessment is necessary or local / general strengthening must be applied or the structure is safe. Similar methods are used after an earthquake event to adjudge if a building suffered critical damage and it is dangerous to be inhabited. D.3 Analytical assessment Analytical assessment methods are supported by the large capabilities of new computers. Usually engineer / analyst set a 3D model of the building. He includes as much and as precise information for materials and geometry of elements, is available from site mapping (pathodology, history), permit or as build drawings. Reliability of the structural analysis, besides the reliability of the engineer/analyst x modelling x software, depends also from the reliability of the data collected. The later parameter introduces uncertainties because i.e. drawings of old buildings are not easy accessible or mapping is difficult, costly, and uncertain. That is why KANEPE prescribes Data Reliability Level, which assigns several safety factors in the analysis. (Karabinis, 2003) Analytical methods can be: method description use elastic lateral force analysis used also for new buildings static behaviour factor q (linear) elastic dynamic multi-modal response spectrum behaviour factor q (linear) preliminary to go to inelastic static inelastic static non-linear static analysis under constant (see below) gravity loads and monotonically increasing horizontal loads (push-over) inelastic dynamic time history analysis Table D-2 types of analytical methods (Karabinis, 2003) dynamic loading inelastic behaviour Inelastic static (pushover) analysis is introduced in codes like FEMA 273/274 and KANEPE, as more reliable analysis in comparison elastic and appropriate for the assessment of existing building response against horizontal loading. It advantages in relation with elastic (linear) methods because identifies plastic hinges, estimates better forces, recognise low capacity regions of the structures, and gives the opportunity to check stresses flow in the structure. D.4 Pushover analysis In general, pushover analysis philosophy is closer to performance-based approach. Main results of a pushover analysis for a building appropriate to depict building s behaviour are: capacity curve (figure) and plastic hinges (place & degree) appendix D-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices Figure D-1 Capacity Curve (Applied technology Council, 1996) This capacity curve is simply a plot of the total lateral seismic shear demand, "V," on the structure, at various increments of loading, against the lateral deflection, of the building at the roof level, under that applied lateral force. (Applied technology Council, 1996) In figure D-1, 3 safety levels (performance levels/target) are depicted: immediate occupancy, life safety, and structural safety. These are levels introduced in US codes (FEMA); they are similar with those in EC-8 (part III) and in Greek KANEPE. The point on the curve that expresses the condition that the capacity of the structure is equal to the demand of the seismic loading is called performance point and quantifies the assessment of structure behaviour. (Applied technology Council, 1996) D.5 Performance levels Similarly, with National Earthquake Hazard Reduction Program (NEHRP), as mentioned in paragraph 5.2.1, the Greek Code for Interventions (KANEPE) is introducing performance targets combining exceedance probability of earthquake loading and performance point of a building bearing structure. [KANEPE, 2009] performance level of bearing structure seismic hazard immediate life safety no collapse exceedance probability T=50yrs occupancy 10% A1 B1 C1 50% A2 B2 C2 Table D-3 Performance targets for assessment and redesign KANEPE (EPPO, 2009) Dimitris Detsis appendix D-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering immediate occupancy life safety Near collapse light damage significant extended no loss of seismic resistance possible loss of seismic total loss of seismic resistance resistance no function corruption function corruption totally unsafe for occupancy limited repairs feasible repairs non-feasible repairs Table D-4 performance level description according to KANEPE The introduction of these levels gives the opportunity to choose the safety level, which is more feasible for the structure in technical and economical way. The following table compares these levels with level of seismic code in Greece (EAK) for new buildings. performance level of bearing structure seismic hazard exceedance immediate occupancy life safety B no collapse C probability T=50yrs A 10% (1) 1,65 1,00 0,70 50% (2) 1,00 0,60 0,45 Table D-5 performance levels & limit states for new buildings KANEPE (EPPO, 2009) D.6 References [D.1] Applied technology Council ATC-40 - Seismic evaluation and retrofit of concrete builidngs. - 1996. - Vol. I. [D.2] EPPO Greek Seismic Code [Book] / ed. (OASP) Earthquake Planning and Protection Organization. - Athens : Ministry of Environment, Planning and Public Works., 2003. - BUILDING CODE - GREEK. [D.3] EPPO Code of Interventions (KANEPE) [Book] / ed. OASP Earthquake Planning and Protection Organization -. - [s.l.] : Ministry of Environment, Planning and Public Works, 2009. - GREEK - BUILDING CODE. [D.4] European Standard Eurocode 8: Design of structures for earthquake resistance - Part 3: Strengthening and repair of buildings. [Book]. - 2003. - pren 1998-1. [D.5] FEMA-356 Prestandard & Commentary for the Seismic Rehabilitation of Buildings [Book]. - Washington DC : Federal Emergency Management Agency, 2000. [D.6] FEMA-389 Primer for Design Professionals Communicating with Owners and Managers of New Buildings on Earthquake Risk [Book]. - [s.l.] : Federal Emergency Management Agency, 2004. [D.7] Karabinis A.I. Assessment of seismic behaviour of r/c constructions Vulnerability and Risk [Conference] // 14th Concrete Conference. - Kos, Greece : [s.n.], 2003. appendix D-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices E. Building typologies - Vulnerability curves This appendix discusses reasoning for the choice of vulnerability curves in the model. (related to chapter 7) This study regards only Greek case and depends on results of large and significant research programs in Greece. The data for vulnerability of Greek buildings were available by the book: Pre-seismic Retrofit of existing buildings edited by National Program for Seismic Retrofit of Existing Buildings, (EPANTYK, 2007) and published Technical Chamber of Greece (T.C.G.). The results are 7 curves for the 7 main typologies of buildings in Greek territory. They do that by coupling damage degree with PGA (occurs) divided by PGA(design). As already mentioned vulnerability, curves for typologies of buildings in Greek territory were available by the results of aforementioned program. They concluded to 7 distinctive typologies for a specific location (and related PGA) depending only in three parameters: date of construction, material of structure and existence of soft ground floor (pilotis). The categorization to 7 typologies and the vulnerability curves are believed to be appropriate for this research and its result; even if at first look seem to be very rough, since several parameters (i.e. height) are missing. The following arguments are supporting this belief. The research program came to this conclusion after more than 10 years of examining, statistically, structural failures after earthquakes in buildings in Greece. At the beginning the parameters were much more than the above three (i.e. morphology of structural system). In the first conclusions, published on 2001, 32 different typologies with distinctive vulnerability are proposed, differentiating according to the above 3 parameters and the height. In the second phase, after the inventory data was available on 2001 and more research, the different typologies were decreased to 7 for the following reasons: 1. No distinction regarding height, because the influence of height is not always the same but depends on characteristics of the specific earthquake. In addition, if height is considered, results are slightly favourable. 2. All the masonry buildings are classified in one category because there are no data for the existence or not of diaphragms. 3. Structures before 1959 and 1959-1985 are in the same category, since there are no big differences in the curves. Height is a very significant parameter for the vulnerability but not for masonry buildings of Greece, which are below 2 floors. For r/c buildings (1) gives a good explanation. General characteristics of the morphology are considered included in the date of construction and material. Other materials and/or special morphologies are a very small percentage of existing stock (3%) and do not influence the result of this study. Actually one Dimitris Detsis appendix E-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering of the most vulnerable morphology item is included, existence or not of ground soft floor (pilotis). Since the result of research depends on statistical data of failures, it can be considered that failures caused by detailing are incorporated. It must be clear that these vulnerability curves are not adequate to be used for an individual building assessment, rather than as decision criterion for retrofit or not an individual building. They are empirical, suitable for risk assessment of large stocks of buildings i.e. in the scale of municipality/prefecture or country. This makes them perfect for this study where the scope is to estimate societal and cost risk of an area (municipality) in order to prioritize resources for seismic risk mitigation. References TCG - Staff Commitee Seismic retrofit existing buildings. Summary of results of 1st phase of research project in G.T.C. [Book]. - Athens : Technical Chamber of Greece, 2001. - GREEK. EPANTYK Pre-seismic Retrofit of existing buildings [Book] / ed. Buildings National Program for Seismic Retrofit of Existing. - [s.l.] : Technical Chamber of Greece (T.C.G.), 2007. - GREEK appendix E-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices F. Description of Greek building stock This appendix gives general description and some pictures of reinforced concrete of building in Greece (related to chapter 5 and 7) F.1 In general The emergence of concrete in the building industry and the introduction of modernism in architecture were the first factors that led towards a new construction system for conventional apartment buildings in urban areas of Greece. The domination of the new system was generalized due to the systematization of the erection process and the typology that corresponded to the emergent need for creation of multi-storey buildings. The load bearing structure consists of a frame of beams, columns, and foundations, made by reinforced concrete. The type of cement, which is used, is PORTLAND C16/20, while St III high yield steel is used as reinforcement. The raw material, in form of dust, is transported to the site. After being mixed with water and a required amount of aggregate, in special units, the viscid material is poured into wooden moulds that are removed after the frame is solidified again. Bricks in the big majority of cases, construct the walls between the concrete frames, both exterior and interior. A single raw of bricks with 9 cm width constructs the interior separations. The exterior ones consist of two parallel layers of brickwork with an internal gap. F.2 In detail For more information regarding Greek multistory reinforced concrete frame buildings please refer to: http://www.world-housing.net/ Some pictures follow: Figure F-2 - construction period - 60 s Figure F-3 - construction period - 70 s Figure F-1 -- construction period - 80 s Dimitris Detsis appendix F-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering Source: Richard Woditsch; Plural- Public and Private spaces of the polykatoikia in Athens; Berlin; 2009 Figure F-4 - construction period - 90 s Figure F-5 - construction period - 00 s F.3 Examples of failures Sources: Elenas A ATHENS EARTHQUAKE OF 7 SEPTEMBER 1999: INTENSITY MEASURES AND OBSERVED DAMAGES [Journal] // ISET Journal of Earthquake Technology. - March 2003. - 1 : Vol. 40. - pp. 77-97. T. P. Tassios, Kostas Syrmakezis, RC Moment Frame Building : Dual System - Frame with Shear Wall, GREECE / HOUSING REPORT / World Housing Encyclopaedia (http://www.worldhousing.net/ ) building collapse 1999 shear wall failure soft ground floor effect soft ground floor effect partial collapse of building appendix F-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices F.4 Some data from this statistical inventory: 3.990.512 buildings all over the country Figure F-6 - Construction time distribution Figure F-7 - Cornerstones introducing new design codes Moreover, in the whole country, 57% are ground level and 30% with one-storey while the 77% are dwellings. Dimitris Detsis appendix F-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering F.5 Especially in Athens 1995 TODAY; 7% not declared; 1% 1985 1995; 13% 1960 1985; 56% BEFORE 1960; 26% BEFORE 1960 1960 1985 1985 1995 1995 TODAY not declared Figure F-8 - Construction time distribution in Athens buildings Furthermore in Athens: 83% are concrete buildings, 61% have 2 or more levels, 95% have pilotis, and 83% are dwellings. Data from Greek National Statistical Authority Buildings Inventory [Online] // GNSA. - http://www.statistics.gr. appendix F-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices G. Retrofit strategies In this appendix some indicative sketches of the main retrofit strategies, as mentioned in chapter 5 are presented here Dimitris Detsis appendix G-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering appendix G-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices H. Field research H.1 Recently published Code of Interventions Code of Interventions KANEPE (EPPO, 2009) is a useful technical handbook, which can help structural engineers to assess the earthquake performance of building and analyse, design and check suitable interventions for strengthening. However, the seminar was very interesting and more than relevant to the subject. In addition in the breaks there was the opportunity to discuss with members of the editorial board and many others colleagues, this gave a good view of what are the intentions now in Greece about the issue of retrofit existing buildings. H.1.1 Key points of the KANEPE: Scope of the code is to enact criteria for the assessment of the capacity of existing buildings and set rules for their earthquake redesign. There is a paragraph referring to legalities and reliabilities of the actors and the users of the interventions projects. Three performance levels are introduced regarding the assessment and the redesign of the existing bearing structure. These are: i. Immediate use after earthquake (=damage limitation EU8-part-3) ii. Life Protection (=significant damage EU8-part-3) iii. Collapse avoidance (=near collapse EU8-part-3) The owner decides upon the performance level of the intervention. With term intervention, we refer to every change of the mechanical characteristics of the building. Repair means to restore the previous condition; strengthen-retrofit means an intervention that improves bearing capacity or ductility of a member or the whole structure. Criteria for decision making about the interventions: i. Satisfaction of earthquake design principles ii. Costs iii. Social needs Four different analysis methods are available. The choice subjects several criteria. Linear elastic, linear dynamic, Non-linear elastic, Non-linear dynamic Several design factors are applied to cover uncertainties in mapping existing situation, modelling the structure, modelling the earthquake action. It is notable the methodology which proposed for assessing the results of survey in the existing building. Dimitris Detsis appendix H-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering H.1.2 Comments On March 2010, Eurocodes are going to be enacted in Greece, that is also refers to Eurocode Part 3 the new European code for seismic assessment of existing buildings. A question is arising concerning the position of KANEPE after the activation of EC8. After reading the paper THE EUROCODE 8 PART 3: THE NEW EUROPEAN CODE FOR THE SEISMIC ASSESSMENT OF EXISTING STRUCTURES, written by Pinto, P.E., to catch up with the new developments in EC8- part 3 it is obvious that the two codes have the same philosophy while the main principles are identical. Actually, members of composing board of KANEPE are also members of the EC composing team. The EC seems to be more descriptive text, giving the general lines and principles of the design, while KANEPE is more detailed. Greeks will ask for the enactment of KANEPE as a National Annex to the EC. Even if there is an effort in chapter 2 of KANEPE, it does not prescribe thoroughly the reliabilities of the actors of such project. Moreover, there is not a strong connection with costs. KANEPE is a very technical, detailed, and strict text providing a very useful methodology tool for the structural engineer, this is what it must have been, and this is what it is. That is why the finalization of the code should NOT be the last step of seismic protection policy but the first. Knowledge of the re-design principles of this code (similar with those of EC8) will help to indentify the relation between 3 performance levels to the cost of intervention. Moreover, gives an indication of the reliabilities of the structural engineer in the project. H.1.3 Conclusions 1. As mentioned also in the previous the completion of the new Code of Interventions or the enactment of EC8 part3 should be only the beginning of process increasing safety against earthquakes. The code is a perfect technical handbook, but process management and general planning should be clarified while cost parameters must integrate in order to succeed in more effective pre seismic risk mitigation policies. 2. KANEPE and EC8 are very similar. One of the most important characteristic is that they introduce 3 performance levels available for the owner to decide. Another point of interest is that they propose 4 types for analysis; this means that they give the opportunity for more accurate assessment of the existing structure s strength and ductility. appendix H-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices H.2 National Program for Seismic Retrofit of Existing Buildings (EPANTYK) EPANTYK is a research & development program running under the funding of Greek Technical Chamber (G.T.C.). The aim is to gather data and information about the building wealth of the country, statistically asses it, estimate their vulnerability, and then evaluate seismic risk, in order to give significant data to the decision makers. The program runs since 1999, consist of engineers and other professionals. The scientific supervisor is Prof. T.P. Tasios. The activities of this program are divided in 3 phases ( 99-01, 01-05 and 05-09). During field research, I had a discussion with NTUA lecturer Emm. Vougioukas member of the program in all its phases, I attended presentation of the result of 3rd phase, found several reports and I was sent the book PRESEISMIC RETROFIT OF EXISTING BUILDING, G.T.C. 2007, which describes the activities of this program. H.2.1 The proposal After the second phase, the program proposed, through G.T.C., the followings to the authorities. With rough and statistical way, the buildings constructed before 1985 might be earthquake vulnerable. This is because: 1. Seismicity of several areas in Greece is larger than estimated by previous earthquake codes 2. Previous codes are below in scientific knowledge and technical accuracy. The request for pre-seismic retrofit of old buildings (80% of building wealth) is backed by the following reasons: 1. Saving resources of after earthquake mitigation 2. Equality among citizens Enormous construction and societal costs as well as psychological, scientific, law and governance problems are major obstacles. Furthermore, the program proposed the following essential actions: administration for existing building retrofit in central government (Ministry of Internal Affairs) pilot program for retrofit public buildings (like schools, hospitals) communication with public (via press and local authorities) local authorities (municipalities or regional authorities) will receive data about the buildings of their territory law changes financial incentives (low interest loans and tax discount etc) legislate new interventions code Dimitris Detsis appendix H-3

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering law and governance modulations for the design and construction processes: engineers qualification, liability and compensation, ways for design check, contractors qualification, upgrading of supervision system, standardization of contracts, consecutive update of project folder ( GREEN BOX ), liabilities allocation among new and old designers/contractors/users, decision making procedure in case of horizontal property, role of building insurance, urban planning changes H.2.2 Approach methodology The methodology, which is proposed by the program, is based on: 1. GIS with the geometrical substratum of the country 2. Inventory of population of year 2001 (national statistical service)/building block 3. Inventory of buildings of year 2000 / building block 4. Descriptive data relative to estimation of vulnerability and seismic risk 5. Data from pilot inventory effort from some towns H.2.3 Seismic risk database software in GIS On 20 th December, I had the opportunity to audit the presentation for a new system, which can be a powerful tool for decision-making. In a GIS substratum, the seismic risk information had been introduced by this way someone can access data about every building block in the state. This tool can help in organising retrofit existing buildings forming a priority policy. Unfortunately, even if the DVD with software is given free to all the municipalities, at this moment I could not have it available. H.2.4 Legal Issues 1. Vague meaning of common recognized technical rules do not provide legal certainty for allocation and attribution of liabilities among project s actors adoption of legal procedures for more strict private project supervision specific legislation for specialization of project and degree of liabilities of its actors clear discrimination of liabilities between initial project s actors and those of intervention project 2. Legislation for the operation of necessary bodies, committees, auditors-experts 3. Minimum safety levels must be determined by the state for: imperative pre-seismic intervention (retrofit) retrofit to meet 4. Specific legislation for assessment and redesign procedure appendix H-4 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices 5. Creation of auditors-experts and civil engineers with retrofit specialization 6. Adaptation of technical code for intervention, which will identify safety levels for existing buildings before and after seismic retrofit (now published KANEPE) 7. Joint ownership is very developed in Greece that is why there is already legislation, which regulates about administration the common property and about entitlements and obligations of joint possessors. 8. Adaptation of specific rules for regulation of retrofit procedure when it comes to adjoining properties 9. Legislation about the maintenance (liabilities and cost distribution) H.2.5 Conclusions Following points outlines some conclusions from the field research: 1. National Program for Earthquake Retrofit of Existing Buildings (EPANTYK) is leading research and development in the field of retrofit existing buildings. Greek Technical Chamber (G.T.C.), Earthquake Planning and Protection Organisation (E.P.P.O.) and other institutions are very interested to boost programs for retrofitting existing buildings. Central government or local seems to be inactive at this moment. 2. Vulnerability and seismic risk assessment elaborated by EPANTYK for the whole country is in a very good stage. At this moment, municipalities have useful GIS tool for decisionmaking. What came up from research is that municipalities do not know what to do with this information. 3. Processing the data of EPANTYK, I conclude that problem is more intense in Athens for and the increased difficult for executing retrofit projects. 4. Research over legal issues and financing ways are also objective of EPANTYK H.3 two reasons, scores high in Vm and R measures national scale (but low in risk /citizen) References EPPO Code of Interventions (KANEPE) [Book] / ed. OASP Earthquake Planning and Protection Organization -. - [s.l.] : Ministry of Environment, Planning and Public Works, 2009. - GREEK - BUILDING CODE. Pinto E.P. THE EUROCODE 8 PART 3: THE NEW EUROPEAN CODE FOR THE SEISMIC ASSESSMENT OF EXISTING STRUCTURES [Journal] // ASIAN JOURNAL OF CIVIL ENGINEERING (BUILDING AND HOUSING). - 2005. - 5 : Vol. 6. T.G.C. Seminar 16 TH December 2009 EPANTYK Pre-seismic Retrofit of existing buildings [Book] / ed. Buildings National Program for Seismic Retrofit of Existing. - [s.l.] : Technical Chamber of Greece (T.C.G.), 2007. - GREEK. TCG - Staff Commitee Seismic retrofit existing buildings. Summary of results of 1st phase of research project in G.T.C. [Book]. - Athens : Technical Chamber of Greece, 2001. - GREEK. th Presentation of results in 20 December 2009 Dimitris Detsis appendix H-5

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering -empty page- appendix H-6 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices I. Principles and scope I.1 Life and wealth protection What is of most important for an engineer is to stay aligned with society needs and wishes. His acts should serve the common good giving the ability for social progress to increase life quality and life expectancy. Engineers who design and construct buildings to cover several people functions, except the functional requirements face also the environmental parameters as boundaries conditions. Among others, a building must withstand natural hazards such as hurricane, earthquakes, floods etc in order to protect people s life but also to ensure their wealth. Life and wealth protection for every human is the main principle for choosing the subject of this thesis. Furthermore, it will be the governing philosophy during the research. I.2 Integration Often occurs technical improvements are late to arrive into the implementation stage. In this cases not all the society, or part of it, gets the opportunity to exploit its advantages. This can happen because either implementation s procedures could not be synchronized or the connection between the technology and implementation is missing. For the second reason integration seems to be the solution. Researchers and scientists must keep in mind that what makes their projects significant is the opportunity to be used in everyday life. From the other side implementation, procedures must be more open to alternations. This is how it works also for the case of a building, the integration between design and construction is crucial to increase value for the end users. The above applies to the chosen subject as well. As it will be described in the following, technical improvements need to enter into realization phase but the implementation environment is not ready yet. I.3 O pen approach In the situation where the implementation could not be synchronized, this is happ ening because an actor is unwilling to lose his privileges. If this attitude is the cause for decreasing life expectancy and social life quality then it is very unacceptable. Open design and stakeholder oriented approach is also one of the principles of this thesis. Research and solution proposal will be elaborated from the point of you of the end user, in order to equally maximize life protection and other benefits. Dimitris Detsis appendix I-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering I.4 Scope The scope of the thesis can be condensed in the followings: provide to the engineering world ideas for integration technological developments with implementation procedures in the case of hazard mitigation, combine knowledge between two different engineering cultures (Dutch and Greek) in ways they confront natural hazards (floods and earthquakes) and give the researcher / author knowledge and experience concerning earthquake engineering and management I.5 Driving parameters Driving parameters for me to work on this subject are: strong belief that pre-mitigation for seismic risk is more efficient than postmitigation, verified complexity of decision making when it comes to risks may abort or delay pre-mitigation projects, integration between structural engineering and construction management, exchange of knowledge and experience between Dutch flood-mitigation methods appendix I-2 14 June 2010

Seismic Risk Mitigation in Greece Translation of Dutch Flood Risk Management Practices Final Report - appendices J. Societal & Individual Seismic Risk Estimator Model (S.I.S.R.E.M.) excel file a CD is attached with the excel file of the developed model and the digital version of this report if you cannot access the contents of thi s CD please contact via e-mail dedimitris@gmail.com Dimitris Detsis appendix J-1

TU Delft / CiTG Design & Construction Processes Section Master Thesis in Building Engineering -last page- appendix J-2 14 June 2010