probabilistic design of coastal flood defences in vietnam

Vietnam has a profound practical experience in the field of flood protection; however, there is a lack of up-to-date knowledge in the field of dike design and flood risk management, espe

Probabilistic Design

of Coastal Flood Defences in Vietnam

Probabilistic Design

of Coastal Flood Defences in Vietnam

Proefschrift

ter verkrijging van de graad van doctor aan de Technische Universiteit Delft,

op gezag van de Rector Magnificus professor ir K.C.A.M Luyben,

voorzitter van het College voor Promoties,

in het openbaar te verdedigen op dinsdag 21 september 2010 om 10.00 uur

door

MAI VAN CONG Master of Science in Hydraulic Engineering geboren te Ninh Binh, Vietnam

Dit proefschrift is goedgekeurd door de promotor:

Prof drs ir J.K Vrijling

Copromotor:

Dr ir P.H.A.J.M van Gelder

Samenstelling promotiecommissie:

Rector Magnificus,

Prof drs ir J.K Vrijling

Dr ir P.H.A.J.M van Gelder

Prof dr ir M.J.F Stive

Lund University, Sweden Newcastle University, UK Formerly Rijkswaterstaat Technical University of Delft Technical University of Delft, reservelid

Copyright © 2010 by Mai Van Cong

All rights reserved No part of this book may be reproduced in any form or by any means including photocopy, without written permission from the copyright holder ISBN 978-90-9025648-1

Printed by: Sieca Repro, Delft, the Netherlands

Cover layout: Pham Quang Dieu

Cover image: Violent ocean wave (source: Ocean Waves Wallpapers)

to my family

S UMMARY

Probabilistic design of coastal flood defences in Vietnam

The year 2005 brought hurricane Katrina to New Orleans (U.S), it was also a historical year for Vietnam because of the disastrous impact of typhoons on sea flood defences and coastal regions In total, eight typhoons hit the Vietnamese coast that year resulting in human casualties and enormous economic damage In response to that, both the central and local governments of Vietnam have paid the highest-ever attention to the rehabilita-tion of sea dike systems and the improvement of the safety of its coastal regions The improvement of the flood defence systems of Vietnam calls for the use of the latest available knowledge at all levels Vietnam has a profound practical experience in the field

of flood protection; however, there is a lack of up-to-date knowledge in the field of dike design and flood risk management, especially regarding to coastal flood defences This study focuses on coastal flood defences in Vietnam and its objectives are to evaluate the current level of safety and reliability, to develop safety standards and design practices and to formulate a proposal for the improvement of the flood defences

To meet the study objectives, the research presented in this thesis has focused on firstly reviewing, developing and widening the application of probabilistic design and state of the art of risk assessment and risk evaluation in the field of flood defences (theory part) Secondly, as a demonstration of knowledge transfer, the proposed approach has been ap-plied to the coastal flood defence system of Nam Dinh province in Vietnam which is con-sidered as a representative coastal system in the country (application part)

The main contents covered in this thesis are summarized below

Chapter 2 gives an overview of the coastal area, existing approaches for flood defence in Vietnam and a description of the case study area The coastal zone in the Red River Delta of Vietnam, where the case study was carried out, is under a potential threat of fluvial flooding, coastal flooding and coastal erosion In the region, sea dikes have been used as a predominant countermeasure with two intended functions, i.e protecting low-lying areas from coastal flooding and reducing risk for the hinterland caused by coastal erosion However, as was also experienced during recent typhoons in the last years, the sea dikes do not seem to function very well and their strength is insufficient to withstand the hydraulic loads with a sufficiently low frequency of occurrence This is due to the underestimation of the hydraulic boundary conditions and imperfections in the design, construction and maintenance of the dike systems A qualitative multi-criteria analysis has been developed for proposing a suitable coastal protection strategy for the study area It is proposed to use a combination of sea dike systems to provide flood protection,

and beach erosion control measures, such as groins and/or sand nourishments to preserve

a shallow foreshore

To have a better understanding of the actual hydraulic load conditions in the coastal area and to derive input information for the reliability analysis, methods for the statisti-cal analysis of extreme hydraulic load conditions have been developed and applied (Chapter 3) These include data management techniques (i.e trend analysis, stationarity tests, seasonality analysis and peaks over threshold analysis); and statistical methods to model the occurrence probabilities of extreme values (the value which is exceeded by the random environmental variable with very small probability p) Hydraulic observations for the coast Vietnam are available for a period of 35 years and on that basis an estimate needs to be derived for a design sea load, i.e waves and storm surge levels, corresponding

to a 100 to 10,000 years return period The methods of Regional Frequency Analysis (RFA) and Mixed Distribution Fitting (MDF) are used in this study for such a sea load

In order to ensure that the statistically predicted values of the interested sea loads will

be as close as possible to the true values, recommendations are made to compare these values with physical hindcast models which are based on historical information of ty-phoon characteristics such as wind speed

Inspection of monthly sea water level data at Nam Dinh coast reveals that there exists a quadratic rising trend, with periodicity of 6 months, 12 months and 224 months in the data The 6-month periodicity reflects well the influence of two monsoons per year to the fluctuation of sea water level; the 12-month periodicity is explained by the influence of yearly typhoons in the region The 224-month periodicity, which is approximately 18.6 years, is almost exactly equal to a tidal circle

Although sea level rise (SLR) is not a focus of this study, based on a detected linear trend of the observed water level data the rate of SLR can be estimated approximately

at 0.68 cm per year at Nam Dinh coast In addition, this study recommends that tion of SLR should be based on not only mathematical extrapolations, but also the long-term nature of sea level fluctuation

estima-The theoretical background of probabilistic design and reliability analysis of flood fences is reviewed in Chapter 4 Models for safety assessment and reliability based design are developed for the specific application to coastal flood defence systems In the safety assessment model, present probabilistic calculation techniques are applied to quantify the failure probability of the system and the system components The effect of the length of the dike system on the system reliability was deployed in the model as well The reliabil-ity-based design model is used to search for an optimal geometry of the dike system given an admissible failure probability

de-Chapter 5 demonstrates the application of the models proposed in the previous chapter

to Vietnam The case study area is the Nam Dinh coastal flood defence system in the Red River delta of Vietnam It is found that the existing coastal flood defence system is less safe than its required safety standards According to the standards the dikes should

be designed to provide a target safety of 1/20 per year However, the total failure ability of the dike system is estimated to be much higher, about 0.15 per year This cor-responds to the occurrence of a failure once in 7 years instead of once in 20 years as in-tended Excessive wave overtopping induced dike failure is the dominant failure mode,

prob-Summary

which likely contributes about 44% to the total failure of the dike system Instability of armour units of the revetments gives an influence of 25% Overflowing and geotechnical instability of inner slope have smaller, but still a considerable contribution (10 and 20 %, respectively) Piping, geotechnical instability of the outer slope and toe foot instability have a negligible influence on the safety A sensitivity analysis shows that wave height and water level are the most important loading parameters The findings are in line with the practical experiences and observations from dike failures in recent years that oc-curred in the Nam Dinh dike system

The application of the reliability based design model shows that if the existing safety standard (1/20 per year) would be applied, the geometry of the current dikes would be insufficient For instance, the crest level of the dike required from the reliability based design is 6.80 meters, whereas the existing crest level of the dike is 5.25 to 5.50 meters

An extensive overview of approaches for risk assessment and risk evaluation is given in Chapter 6 In general, risk is defined as the product of probability of occurrence and con-sequence of disasters to be considered Regarding flood risk in this study, the conse-quences are considered in terms of loss of life (individual risk† and societal risk‡) and economic damage (economic risk) A central question in risk management concerns the acceptance of risk by the public and the decision makers Presently, the acceptable flood-ing risk is judged from two points of view which relate to the acceptance of inundation consequences The first is acceptable risk regarding loss of life, which is under judgement

of society and individuals The second is viewed under an economic cost benefit balance Chapter 7 and Chapter 8 present applications of the theory given in Chapter 6 to the Vietnamese situation In Chapter 7, an approach for estimation of the level of acceptable flood risk for Vietnam is proposed The approach is based on earlier framework of Vri-jling (2000) in which acceptable risk level is based on characteristics of the activity (benefit, voluntariness) and baseline mortality statistics in a country In this assessment, first the mortality due to natural causes and daily activities such as traffic is involved Based on Vietnamese statistics, baseline mortality has been determined including the risks of traffic and living in a flood-prone area in Vietnam

An FN-curve for flooding in Vietnam is constructed showing the probability of accidents with fatalities Adapting the risk evaluation framework of Vrijling (2000), the national level of acceptable risk of Vietnam is established Evaluation of the present floods on the national scale shows that acceptable societal risk of Vietnam is five to ten times higher than in the Netherlands

To obtain a risk limit for the Nam Dinh coastal region, the nationally acceptable risk has

to be distributed over all the regions in one country A societal risk limit for Nam Dinh

is obtained and it is found that the current level of flood risk exceeds this limit by far According to the limit the flood risk is acceptable if the low lying coastal region of Nam

Dinh is protected to a safety standard with exceedance probabilities lower than 1/150 per year

The optimal safety is evaluated based on an economic approach for risk based design in Chapter 8 Historical data on economic loss due to floods in Vietnam is available since

1970 The distribution of these losses can best be approximated by a lognormal tion and the yearly expected value of the damage equals approximately $US 589x106 per year This value is comparable with the actual flood damage of Vietnam during the last years (over the the last 10 years the total flood damages on average are about 1.0% of the GDP of Vietnam per year)

distribu-Economic risk analysis of the Nam Dinh dike ring shows that the safety standard of 1/50 years should be applied to the present economic situation (Situation 1) Because of the fast economic growth and accumulated wealth in the future (Situation 2), a safety stan-dard of 1/100 is proposed

The societal risk of flooding in Nam Dinh as well as in Vietnam appears to be able according to some existing risk limits that have been referred to literature These results indicate the necessity of more societal discussion and further studies on the ac-ceptable level of flood risk in Vietnam The decision has to be made if the current risks are acceptable or additional risk reducing measures are necessary The methods and re-sults presented in this thesis provide an important input background for making these decisions

unaccept-Delft, 2010 July 29

Mai Van Cong

Summary

Probabilistisch ontwerp van waterkeringen langs de kust van Vietnam

In het jaar 2005 werd New Orleans (Verenigde Staten) getroffen door orkaan Katrina Het was ook een historisch jaar voor Vietnam vanwege de catastrofale impact van ty-foons op de kustverdediging en kustgebieden In totaal hebben acht tyfoons de Vietna-mese kust bereikt Deze resulteerden in menselijke slachtoffers en enorme economische schade Als repsons op deze gebeurtenissen hebben de lokale en centrale overheden van Vietnam zeer veel aandacht besteed aan het herstel van de waterkeringen langs de kust

en de verbetering van de veiligheid van de kustgebieden

Voor de verbetering van het waterkeringsysteem van Vietnam is de meest recente kennis

op verschillende niveaus benodigd Vietnam heeft uitgebreide praktijkervaring op het bied van hoogwaterbescherming Echter, er is een gebrek aan up-to-date kennis over dijkontwerp en de beheersing van overstromingsrisico’s, specifiek voor kustgebieden Deze studie richt zich op de waterkeringen in de kustgebieden van Vietnam De doelen betref-fen het evalueren van het huidige veiligheidsniveau en de betrouwbaarheid van de water-keringen, het ontwikkelen van veiligheidsnormen en ontwerprichtlijnen en het formuleren van een voorstel voor de verbetering van de waterkeringen

ge-Om deze doelen te bereiken heeft het onderzoek zich ten eerste gericht op het evalueren, ontwikkelen en verbreden van de toepassing van probabilistisch ontwerpen en risicoanaly-

se voor waterkeringen (het theorie deel).Vervolgens is, mede in het kader van dracht, de voorgestelde aanpak toegepast op het waterkeringssysteem langs de kust van

kennisover-de provincie Nam Dinh in Vietnam Dit gebied is representatief voor het kustsysteem in Vietnam (het toepassings deel)

De belangrijkste onderwerpen die in dit proefschrift zijn behandeld zijn hieronder mengevat

sa-Hoofdstuk 2 geeft een overzicht van het kustgebied, de bestaande wijze van bescherming in Vietnam en een overzicht van het case studie gebied De kustzone in de delta van de Rode Rivier in Vietnam, waarvoor de case studie is uitgevoerd, wordt be-dreigd door rivieroverstromingen, kustoverstromingen en kusterosie In het gebied hebben waterkeringen langs de kust twee functies, namelijk het beschermen van laaggelegen ge-bieden tegen overstroming en het beperken van de risico’s voor het achterland door kus-terosie Echter, tijdens recente tyfoons is gebleken dat de waterkeringen langs de kust niet goed functioneren De sterkte van de keringen is zelfs onvoldoende om hydraulische belastingen die relatief vaak voorkomen te weerstaan Dit komt (mede) door de onder-schatting van de hydraulische belastingen in de huidige praktijk en imperfecties in het ontwerp, de constructie en het beheer en onderhoud van dijksystemen Een kwalitatieve multi-criteria analyse is ontwikkeld om geschikte strategieën voor kustverdediging van het gebied te bepalen Er wordt voorgesteld om een combinatie van zeedijken voor be-

♣ This “Samenvatting” is translated by Bas Jonkman and Wim Kanning (TUDelft) from its lish version

Eng-scherming tegen overstromingen, en kusterosie beperkende maatregelen, zoals hoofden en zandsuppleties toe te passen om zo een flauwe vooroever the handhaven Methoden voor de statistische analyse van extreme hydraulische belastingen in het kust-gebied zijn ontwikkeld en toegepast (hoofdstuk 3) Hiermee ontstaat een beter begrip van

strand-de daadwerkelijke hydraulische belastingen en komt informatie beschikbaar waarmee een betrouwbaarheidsanalyse kan worden uitgevoerd De beschouwde methoden betreffen technieken voor data beheer (trend analyse, stationariteits testen, seizoenscyclus analyse

en peaks-over-threshold analyse) en statistische methoden om de kans op het voorkomen van een extreme waarde (de waarde die wordt overschreden door de willekeurige omge-vingsvariabele met een zeer kleine kans) te bepalen Hydraulische observaties voor de kust van Vietnam zijn beschikbaar over een periode van 35 jaar Op basis daarvan dient een schatting gemaakt te worden van een ontwerp belasting, bestaande uit golven en stormopzet, die overeen komt met een herhalingstijd van 100 tot 10.000 jaar De metho-den Regionale Frequentie Analyse (RFA) en het fitten met Gemengde Verdelingen zijn gebruikt voor het afleiden van de hydraulische belastingen voor het kustgebied Om zeker

te stellen dat de statistisch voorspelde waarden van de relevante belastingen dicht bij de daadwerkelijke waarden liggen, zijn aanbevelingen gedaan om deze waarden te vergelij-ken met de resultaten van fysische modelevaluaties van historische gebeurtenissen Deze zijn gebaseerd op historische kenmerken van tyfoons, zoals de windsnelheid

Hoewel de analyse van zeespiegelstijging niet de focus van deze studie betreft, blijkt dat

de zeespiegelstijging lineair verloopt Op basis van de beschikbare gegevens is de absolute zeespiegelstijging ingeschat op 0,68cm per jaar voor de kust van Nam Dinh Daarnaast is aanbevolen om de inschatting van zeespiegelstijging niet alleen te baseren op wiskundige extrapolatie van gegevens, maar ook op de langjarige cycli die het zeespiegelniveau bepa-len

De theoretische achtergrond van het probabilistisch ontwerp van waterkeringen is schouwd in hoofdstuk 4 Modellen voor de veiligheidstoetsing en op betrouwbaarheid ge-baseerd ontwerp zijn ontwikkeld voor de specifieke toepassing voor waterkeringssystemen langs de kust In het model voor veiligheidstoetsing zijn probabilistische rekentechnieken gebruikt om de kans op het falen van het systeem en systeemcomponenten te kwantifice-ren Het effect van de lengte van het dijkensysteem op de betrouwbaarheid van het sys-teem is in het model ook meegenomen Het model voor op betrouwbaarheid gebaseerd ontwerp is gebruikt om een optimale geometrie van het dijksysteem af te leiden uitgaan-

be-de van een toelaatbare faalkans

Hoofdstuk 5 demonstreert de toepassing voor Vietnam van de modellen die in het vorige hoofdstuk zijn behandeld Dit is gedaan in een case studie voor de Rode rivier delta in

de provincie Nam Dinh in Vietnam Het blijkt dat het waterkeringsysteem langs de kust minder veilig is dan wordt beoogd met de bestaande ontwerpnormen Volgens deze nor-men zouden de dijken ontworpen moeten worden om een veiligheid van 1/20 per jaar te bieden Echter, de totale faalkans van het dijken system wordt hoger geschat, op onge-veer 0,15 per jaar Dit komt overeen met het optreden van een faalgeval gemiddeld eens per 7 jaar, in plaats van de beoogde faalkans van eens per 20 jaar Falen door excessieve golfoverslag is het dominante faalmechanisme en dit draagt voor ongeveer 44% bij aan de faalkans van het totale systeem Instabiliteit van steenbekledingen draagt 25% bij Over-loop en geotechnische instabiliteit van het binnentalud hebben een kleinere, maar nog

Summary

steeds aanzienlijke bijdrage aan de faalkans (10%, respectievelijk 20%) Piping en technische instabiliteit van het buitenland talud en de dijkteen hebben een verwaarloos-bare bijdrage Uit een gevoeligheidsanalyse blijkt dat de golfhoogte en de waterstand de belangrijkste belasting variabelen zijn

geo-De bevindingen liggen in lijn met praktijkervaringen en observaties betreffende recente dijkdoorbraken in het dijkensysteem van Nam Dinh De toepassing van het model voor

op betrouwbaarheid gebaseerd ontwerp laat zien dat de afmetingen van de bestaande dijken onvoldoende zijn om te voldoen aan de huidige veiligheidsnorm (1/20 per jaar) Het blijkt bijvoorbeeld uit de betrouwbaarheidsgebaseerd ontwerp dat de benodigde kru-inhoogte 6.80 meter moet zijn, terwijl de bestaande kruinhoogte 5.25 tot 5.50 meter be-draagt

Een uitgebreid overzicht van methoden voor risicoanalyse en risico evaluatie is gegeven in hoofdstuk 6 Over het algemeen wordt risico gedefinieerd als het product van de kans op het optreden van rampen en de gevolgen daarvan In deze studie is het overstromingsrisi-

co beschouwd in termen van slachtofferrisico (individueel** en maatschappelijk risico††) en economische schade (economisch risico) Een centrale vraag in risico management betreft

de aanvaardbaarheid van risico’s voor het publiek en beleidsmakers De heid van overstromingsrisico’s wordt over het algemeen vanuit twee gezichtspunten be-oordeeld Het eerste betreft de aanvaardbaarheid van slachtofferrisico’s, zowel voor indi-viduen als grote groepen Het tweede gezichtspunt heeft betrekking op een economische balans van kosten en baten

aanvaardbaar-In hoofdstuk 7 en 8 is de in hoofdstuk 6 gepresenteerde theorie toegepast op Vietnam aanvaardbaar-In hoofdstuk 7 is een aanpak voor de bepaling van het aanvaardbare niveau van overstro-mingsrisico’s voor Vietnam voorgesteld De aanpak is gebaseerd op een eerder raamwerk van Vrijling (2000) waarin het acceptabele risico niveau wordt bepaald op basis van de kenmerken van de activiteit (baten, vrijwilligheid) en statistieken voor sterfte door diver-

se oorzaken in een land Hierbij wordt allereerst de sterfte door natuurlijke oorzaken en reguliere activiteiten, zoals deelname aan het verkeer, beschouwd Op basis van beschik-bare statistieken, is de basis sterftekans voor Vietnam bepaald inclusief de risico’s die samenhangen met het verkeer en het leven in overstromingsgevoelige gebieden in Viet-nam

Een FN curve voor overstromingen in Vietnam is afgeleid en deze toont de kans op beurtenissen met bijbehorende aantallen slachtoffers Het raamwerk voor risico evaluatie van Vrijling (2000) is aangepast en het nationale acceptabele risiconiveau voor Vietnam

ge-is afgeleid Evaluatie van het overstromingsrge-isico op nationale schaal laat zien dat het acceptabele maatschappelijk risiconiveau voor Vietnam vijf tot tien maal hoger is dan voor Nederland

Om een risiconorm vast te stellen voor het kustgebied van Nam Dinh, is het nationaal aanvaardbare risico verdeeld over alle regio’s in het land Hiermee wordt een risiconorm

** Individueel risico: de kans (per jaar) om om het leven te komen op een zekere locatie uitgaande van permanente aanwezigheid van de bevolking

†† Maatschappelijk risico: de overschrijdingskans (per jaar) van een ongeval met meerdere offers Maatschappelijk risico wordt vaak afgebeeld in een FN curve Deze toont de kans per jaar

slacht-op ongevallen met N of meer slachtoffers

voor Nam Dinh verkregen en het blijkt dat het huidige overstromingsrisico ver boven de norm ligt Volgens de norm is het overstromingsrisico aanvaardbaar als laaggelegen kust-gebieden een veiligheidsnorm krijgen van ten minste 1/150 per jaar

Het optimale veiligheidsniveau is geëvalueerd in hoofdstuk 8 op basis van een sche benadering voor risico-gebaseerd ontwerpen Historische gegevens over economische schade door overstromingen in Vietnam zijn beschikbaar sinds 1970 De statistische ver-deling van deze schades kan het best benaderd worden met een lognormale verdeling De karakteristieke waarde van de jaarlijkse schade is ongeveer gelijk aan $US 589x106 per jaar Deze waarde is vergelijkbaar met de daadwerkelijk opgetreden schade door over-stromingen in Vietnam in recente jaren Gedurende de laatste tien jaren bedroegen de schades door overstroming gemiddeld 1,5% van het bruto nationaal product van Viet-nam

economi-Een economische risicoanalyse (optimalisatie) voor het dijkensysteem van Nam Dinh laat zien dat een veiligheidsnorm van 1/50 per jaar gekozen zou moeten worden uitgaande van de huidige economische situatie Rekening houdend met snelle economische ontwikke-ling in de toekomst wordt een veiligheidsnorm van 1/100 per jaar voorgesteld

Het maatschappelijk risico door overstromingen in Nam Dinh en de rest van Vietnam lijkt niet aanvaardbaar volgens risiconormen die zijn afgeleid op basis van de literatuur Deze resultaten wijzen op de noodzaak van meer maatschappelijke discussie over en ver-der onderzoek naar het aanvaardbare niveau van overstromingsrisico’s in Vietnam Be-sluitvorming is nodig over de aanvaardbaarheid van de bestaande risico’s en de noodzaak van aanvullende maatregelen De methoden en resultaten die zijn gepresenteerd in dit proefschrift leveren belangrijke achtergrond informatie aan om deze besluiten te nemen Delft, 29 juli 2010

Mai Van Cong

C ONTENTS

SUMMARY I

CONTENTS 9

CHAPTER 1INTRODUCTION 15

1.1 General background on flood defences 15

1.2 Developments in the design and safety assessment of flood defences 16

1.3 Catalysts in new policy towards flood risk 18

1.4 Additional stimulus of this research 19

1.5 Probabilistic vs deterministic approach 19

1.6 Fundamentals of this study 21

1.6.1 Problem definition 21

1.6.2 Study objectives and approach 21

1.7 Scientific and social relevance 23

1.7.1 Scientific relevance 23

1.7.2 Social relevance 24

1.8 Structure of this thesis 24

CHAPTER 2COASTAL MANAGEMENT AND FLOOD DEFENCES IN VIETNAM 27

2.1 Flood defences in Vietnam: current situation and design practice 27

2.2 Coastal protection in the Red River Delta: current strategies 29

2.2.1 Description of study area 29

2.2.2 Causes of coastal erosion 32

2.2.3 Historical information on coastline changes 34

2.2.4 The year 2005 of typhoons and its consequences 36

2.3 Hydrodynamics in the gulf of Tonkin 37

2.4 Longshore sediment transport and coastline changes 38

2.5 Actual safety of the coastal protections 40

2.6 Discussion 42

2.7 Conclusions 45

CHAPTER 3ESTABLISHING THE COASTAL HYDRAULIC CONDITIONS 47

3.1 Introduction 47

3.2 Regional frequency analysis 47

3.3 Data management methods in extreme value analysis 49

3.3.1 The Annual Maxima (Gumbel) Method 49

3.3.2 Peak-over-threshold in modelling extreme value data 50

3.3.3 Trend analysis, stationarity test and detection of seasonality and natural fluctuation 51

3.4 Statistical background use in RFA 52

3.4.1 Definitions, steps in RFA and site discordance detection 52

3.4.2 L-Moment statistics 53

3.4.3 Discordance measure 55

3.4.4 Robust distances 55

3.5 RFA of extreme sea water level along the Vietnamese East sea coasts 57

3.5.1 Extreme sea water level data 57

3.5.2 Regional frequency analysis of the sea data 58

3.5.3 Discussion on the application of RFA 61

3.6 Statistical model of extreme sea water level at Nam Dinh coast 62

3.6.1 Introduction 62

3.6.2 Data inspections: rising trend, periodicity and sea level rise 63

3.6.3 Annual maxima Gumbel to model the extreme 65

3.6.4 Mixed distribution to model the extreme 66

3.6.5 Summary 69

3.7 Combined statistics and physics 69

3.7.1 Conceptual model and steps 69

3.7.2 Example of implementation 70

3.8 Discussion 73

CHAPTER 4RELIABILITY-BASED DESIGN:THEORY 75

4.1 Introduction 75

4.2 Definition of an engineering system 75

4.3 Reliability of a system component 77

4.3.1 Non linear Z-function and design values 81

4.3.2 Non normally distributed basic variables 83

4.4 Analysis of a system 83

4.4.1 Basic systems 84

4.5 Reliability of a system 86

4.6 Length effects and system reliability 89

4.6.1 Hypothetical case: a dike system comprises one-type cross section 91

4.6.2 Dike system comprises multi dike sections 96

4.6.3 Correlation distance/scale of fluctuation 98

4.7 Discussion 99

Contents

4.7.1 Application model 1: Reliability safety assessment 99

4.7.2 Application model 2: Reliability based optimal design 100

4.7.3 On the probabilistic models and reliability calculation methods 101

CHAPTER 5RELIABILITY-BASED DESIGN OF COASTAL FLOOD DEFENCES IN VIETNAM 109 5.1 Introduction 109

5.2 Description of Nam Dinh coastal flood defences 110

5.3 Possible failure mechanisms of the sea dike systems 112

5.4 Physical process-based limit state equations of failure mechanisms 114

5.5 Relibility analysis of the Nam Dinh sea dike system 119

5.5.1 Safety assessment 119

5.5.2 Fault tree analysis and system failure probability 120

5.5.3 Sensitivity analysis of stochastic variables 122

5.5.4 Section length effects 124

5.5.5 System length effects 129

5.5.6 Total failure probability of Hai Hau dike system 131

5.5.7 Translation of system safety to design frequency of cross sections 132

5.5.8 Total failure probability of Nam Dinh coastal flood defence system 134

5.5.9 Reliability-based design 135

5.6 Discussion 137

CHAPTER 6RISK ANALYSIS CONCEPTS AND RISK-BASED DESIGN: THEORY REVIEW 141

6.1 Historical development of risk analysis and risk management 141

6.2 Recent developments in risk analysis 143

6.2.1 Overview 143

6.2.2 What is risk? Risk definition and presentation 144

6.2.3 Discussion of risk measures for loss of life 146

6.2.4 Damage (economic) risk 149

6.2.5 Flood risk 149

6.2.6 Risk perception 150

6.2.7 Risk acceptance 151

6.3 Development of flood risk policy in the Dutch situation 152

6.3.1 Overview 152

6.3.2 Framework for determination of acceptable risk due to floods 153

6.3.3 Individual risk 154

6.3.4 Societal risk 155

6.3.5 Estimation of potential loss of life due to floods 158

6.3.6 Economic optimisation 160

6.4 Discussion 162

6.4.1 Risk related to loss of life 162

6.4.2 Economic risk based approach 163

6.4.3 An adapted economic risk-based framework for flood defences in Vietnam 164

CHAPTER 7RISK EVALUATION AND ACCEPTABLE RISK IN THE VIETNAMESE SITUATION 167

7.1 Introduction 167

7.2 Assessment of actual risks and present safety standard 167

7.2.1 Present risks due to (all possible) natural causes 168

7.2.2 National acceptable risk levels: Safety standard for Vietnam 168

7.2.3 Present risk due to traffic accidents 169

7.2.4 Historical loss of life due to storms and floods in Vietnam 170

7.2.5 Discussion 172

7.3 Loss of life due to flooding in Vietnam 173

7.3.1 Individual risk for coastal flooding in Vietnam 173

7.3.2 Societal risk due to flood on a country level 174

7.4 Case study: analysis of flood risk in Nam Dinh coastal zone 179

7.4.1 Loss of life estimates for Nam Dinh coastal zone 179

7.4.2 Acceptable flood risk for Nam Dinh coastal zone 184

7.4.3 Discussion 186

CHAPTER 8APPLICATION OF ECONOMIC RISK-BASED DESIGN 187

8.1 Economic losses of Vietnam due to flooding 187

8.2 Boundary conditions 188

8.2.1 Cost estimates 188

8.2.2 Sea water levels 189

8.2.3 Dike height in relation to exceedance frequency of water level 190

8.3 Economic optimisation of Nam Dinh sea flood defences 191

8.3.1 Cost optimal function 191

8.3.2 General assumptions 192

8.3.3 Determination of potential economic risk 193

8.3.4 Summary of the results 193

8.3.5 Sensitivity analysis 195

8.4 Extension of the van Dantzig model 197

8.4.1 Basic case: van Dantzig model (1956) 197

8.4.2 Protected area with inland slope 200

8.4.3 Sensitivity analysis 204

8.5 Discussion 205

CHAPTER 9CONCLUSIONS AND RECOMMENDATIONS 207

9.1 Conclusions 207

9.1.1 General 207

9.1.2 Establishment of hydraulic boundary conditions for coastal areas 208

Contents

9.1.3 Investigation and development of safety assessment and reliability based design 209

9.1.4 Risk based design with a special application to the Vietnamese situation 210

9.1.5 Application to the Vietnamese coastal flood defences 210

9.2 Recommendations 212

9.2.1 Reliability and safety assessments 212

9.2.2 Acceptable risk 213

9.2.3 Establishment of sea boundary conditions 213

9.2.4 The Nam Dinh sea defences and optimal safety of Nam Dinh coastal zone 214

9.2.5 Coastal protection strategies for the Red River delta of Vietnam 215

9.2.6 Vietnam flood risk and flood defence strategies 216

REFERENCES 217

LIST OF FIGURES 227

LIST OF TABLES 231

APPENDICES 233

Appendix 1 List of scientific papers 233

Appendix 2 Statistical description of the random variables used in reliability analyses 235

Appendix 3 Overview of population data of Nam Dinh 236

LIST OF NOTATIONS 239

ACKNOWLEDGEMENTS 243

ABOUT THE AUTHOR 245

Chapter 1

Coastal and fluvial areas bring enormous benefits for human life, however both areas are liable to potential risk the sea and rivers, i.e storm waves and surges, tsunamis, high river discharges and consequently floods To avoid flooding risks while living in these ar-eas our ancestors adapted to nature recognisably In coastal areas human settlements were in the higher natural areas or on man made hills, named "mounds" In the river ar-eas people lived on the natural higher ground with lower area were being used for grow-ing food As a result, the regular floods had little negative effects and even deposited fer-tile silt on the land, which approximately enabled the land to keep pace with the natu-rally rising sea level

The rise in population meant that increasing numbers of lower lying areas were taken into use At the same time, provisions were put into place in relation to agriculture to drain the land; wells were made for drinking water and peat was dug up in many places for fuel and salt production The consequence was a fall in the level of the ground sur-face, as a result of which flooding became a greater problem

In addition, a rise in population is often accompanied by a rise in human actives in the flood-prone areas, which often results in an increase of socio-, cultural- and economic- values This means more value and life may be at risk due to flooding and need to be protected

In response, dikes were constructed Initially, they only supplemented the natural heights; in the rivers area for instance, perpendicular to the bank walls to redirect river water flowing outside the banks along populated areas to lower-lying back lands In the coastal area the influence of the sea increased steadily due to sea level rise and land sub-sidence, which means that, more and more lowlands had to be protected by dikes (Ton-neijck & Weijers 2008)

Generally, almost all low-lying coastal and/or fluvial land around the world is densely populated Many large-scale urban expansions are concentrated in low-lying regions where below mean sea level (MSL) In time, not only the protected areas increase in size,

but also the level of required protection increases The value of property logically creases and the number of human lives threatened by flooding increases as well The con-sequences of a flood will, therefore, be much more serious than before as the inundation depth and the protected values of the area increased during the centuries

in-In summary, structural problems are: a densely populated, highly developed, but lying areas, where flooding could lead to loss of human life, economic, cultural and less tangible values and disruption of a society There is no other way than ensuring that we secure our residential areas prudently Therefore, flood defences are needed

Later, when the population and the need for space increased the development of dikes began Dikes covered protected areas alongside rives and seas The water flowing down the river would be diverted to a place where it would cause less damage In time, as the need for more space increased, the amount of protected area grew

The first dikes were designed based on experience and skill of the builder The experience and skill was gathered in practice Practice was to build the dike at the highest known water level, which was the so-called design water level, plus a certain additional free-board, a half or one meter was applied worldwide This design water level and the con-struction techniques used had no scientific base This high water level in the river is mainly caused by heavy rain upstream, while along the sea coasts, this is often caused by surges during storms

In the last century, mathematical and statistical knowledge improved Combined with the introduction in practice of fluid and soil mechanics the approach for dike design be-came more and more scientific The hydraulic load on a flood defence could be predicted more accurate and the strength of the flood defences could be calculated, always based

on extreme value extrapolation of the existing situation

In the Netherlands, after the disaster in 1953 a statistical approach to the storm surge levels was chosen and an extrapolated storm surge level formed the basis for dike design

It was also described how the Delta Committee optimised these safety levels, which were expressed in terms of the return period of high water levels which must be withstood by the primary dike system (van Dantzig, 1956)

1.2 Developments in the design and safety assessment of flood defences

Since 1980, the development and the application of reliability theory made it possible to assess the flooding risks taking into account the multiple failure mechanisms of a dike or

a structure Dutch hydraulic engineers were among the first to apply this theory in the practical design of structures Reliability models were first used during the design and the construction of a storm surge barrier in the Easternscheldt in 1976 and later in the design of a storm surge barrier in the Nieuwe Waterweg

In approximately 1979 a project was started to apply the probabilistic methods to the design of dikes in general (Vrijling, 2001) The development of a complete approach to water defense systems took a considerable time (Tonneijck & Weijers, 2008) Recently, the approach was tested applied on many Dutch polders or dike rings The probability of flooding of these polders was calculated and insight in the weak spots was gained Re-sults can be viewed in detail e.g in Vrijling et al (2001), Jonkman et al (2008) and FLORIS reports6

Actually, the results of the calculations stimulated political debate about whether the present safety level of the water defense system is still sufficient in the Netherlands This question should be posed because the economic activity in the protected areas has grown considerably since the defence systems were constructed Moreover, ambitious private and public investments, particularly in infrastructure, are planned The national econ-omy has and will become more vulnerable to flooding In addition, the safety of a grow-ing number of inhabitants is at stake These types of situations and discussions are not only occurring in the Netherlands, but also in many other flood prone countries Gener-ally, the number of discussions on how safe is safe enough has been of increasing atten-tion nowadays worldwide

Probabilistic design with risk based design concepts are considered the most modern proaches in the field of flood defences Advantages of the methods are that allows de-signer to take into account the uncertainties of the input parameters as random vari-ables; and to describe the flood defence as a system, including various structural compo-nents and its protected area Moreover, for each system component various possible fail-ure modes can be considered These all help to determine the true probability of flooding

ap-of a protected area and to judge its acceptability in a view ap-of the consequences ap-of the protected area Thus, probabilistic approach is apparently an essential tool to analyse the flood defence systems

Although the probabilistic approach has been increasingly proposed and applied in the last years in the West (see e.g the concept, method and application in Bakker & Vrijling

1980; Vrijling et al 1998; Hall et al 2000; Oumeraci et al 2001 and Voortman 2002;),

application of these methods in developing countries is still limited

In Vietnam, the approach has only recently been introduced as a new modern design methodology Still very few applications of the approach have been done (Mai Van et al 2007b) Thus, widening the applications of the method to the Vietnamese situation of flood defences and coastal protection is therefore necessary The methods can also be ap-plied to cases of other developing countries

6 FLORIS: Floodrisk and Safety assessment of primary flood defences in the Netherlands

1.3 CATALYSTS IN NEW POLICY TOWARDS FLOOD RISK

One of the tasks of human civilisation is to protect individual members, groups of people and their possessions against natural and man-made hazards to an acceptable extent The extent of the protection in historical cases was mostly decided after the occurrence

of the hazard which had shown the consequences The modern probabilistic approach aims to give protection when the risks are assessed to be high Risk is defined as the probability of a disaster e.g a flood related to the consequences As long as the modern approach is not firmly embedded in society, the idea of acceptable risk may be quite suddenly influenced by a single spectacular accident or incident

For example in the Dutch case, after the sea flood disaster happened in 1953, the Dutch society got a severe wake up call This event showed the vulnerability of the country and the lack of science to base dike design on After this event the implementation of sophis-ticated techniques was accelerated The Delta Committee (a committee that advised on the strategy needed to prevent disasters like the one that occurred in 1953), proposed a new way of looking at the safety of the Netherlands against flooding This committee was aware of the fact that not all the knowledge needed to implement this new approach was available at that point, but the new way of thinking enabled the development of the scientific approach of the flood defences One of the important changes was the develop-ment of probabilistic techniques in hydraulic engineering Using statistical techniques for the hydraulic load was the starting point The realisation that the strength of a flood de-fence is a parameter with an average and a standard deviation changed the way of de-signing dikes as well

In 2005, hurricane Katrina, which hit New Orleans, induced a new policy and attention

of the U.S to the sea defences Moreover, the influence of Katrina on flood risk ment policy is not only valid for the U.S, but also for many other flood prone countries

manage-In Netherlands, the New Orleans flood was critically reviewed to understand what had happened (Kok et al., 2007) Lessons learnt partially stimulated the establishment of a new Dutch Delta Committee

In Europe, more attentions and efforts are paid to reducing flood risk Implementation of the EU floods Directive is under way Government taskforces are preparing emergency as well as longterm management for flood An example would be the FLOODSite project7,

an integrated project that involved of 25 countries with 35 water related institutes and delivered advances in several areas of direct relevance to the three main activities of the Directive, including:

- preliminary flood risk assessment;

- the preparation of flood risk maps;

- the preparation (and implementation) of flood risk management plans

Preferences for other types of measures i.e insurance, evacuation, using compartment dikes and room for water are also under discussion in Europe

7 FLOODSite: EU funded project under the Framework 7 (2004-2009)

1.4 Additional stimulus of this research

In Vietnam, the year 2005 was considered a historical year because of the disastrous pact of typhoons on its coastal regions In total eight typhoons hit the Vietnamese coasts

im-in that year resultim-ing im-in human casualties and large economic damage Especially, phoon Damrey was of exceptional strength and was one of the heaviest typhoons of the last 5 decades Damrey occurred in September 2005 in northern Vietnam and caused ap-proximately more that 28 km of sea dike breaches at different sections along coastline of Nam Dinh and Thanh Hoa province The typhoon affected more than 1 million people and claimed serious economic loss of over 500 Million USD (DMWG, 2005) As a wake-

ty-up call for Vietnam, its government has paid the highest ever attention to rehabilitating the sea dike systems and improving safety of the coastal regions

Improvement of the flood defence system in Vietnam calls for the use of present available knowledge on all levels Vietnam has profound practical experience in the fields of flood protection; however, there is lack of recent theoretical knowledge in the fields of dike de-sign, safety and reliability, risk -based approaches in flood defences Therefore, the trans-fer of these up-to-date knowledge and experiences was strongly recommended

(DWW/RWS, 1996; Vrijling et al., 2000; Mai et al., 2006g) to Vietnam

In an initial attempt to improve safety for the coastal low-lying regions and rehabilitate the sea dike systems on the long run, a huge sea dike research program has been initiated

by the Ministry of Agriculture and Rural Development (MARD) in 2006 The program runs from 2007 to 2012 covering the whole of Vietnam The main tasks of the program are an assessment of the present situation of the sea defence regarding the safety and the establishment of new safety standards and developing a master plan for protecting the low-lying coastal regions The theoretical background and technical knowledge in design-ing of sea flood defences and establishing sea boundary conditions are needed Expected products of the program will be technical reports, covering all theoretical background and practical examples, safety standards and a master plan of coastal sea protections; boundary condition books, which will be used in sea defences design

This research, amongst other research aims, is a part of knowledge transfer for the namese situation in flood defences, in general, and sea flood defences, in particular

The conventional design method in civil engineering design is a deterministic approach The basis of the deterministic approach is the so-called design values for the loads and the strength parameters Loads for instance are the design water level and the design significant wave height Using design rules according to codes and standards it is possible

to determine the dimensional geometrical parameters of the civil engineering structures i.e buidings, bridges, tunnels, dams, dikes and storm surge barriers These design rules are, in general, based on limit states of the structure’s elements

In the deterministic approaches it is assumed that the structure is safe when the margin between the design value of the load and the characteristic value of the strength is large

enough for all limit states of all elements Therefore the safety level of a structured tem is not explicitly known (CUR 141, 1990)

sys-Probabilistic design with reliability- and risk- based design concepts have been ingly proposed and applied in the fields of civil engineering and flood defences during the last years A fundamental difference with the deterministic approach is that the probabil-istic design methods are based on an acceptable frequency or probability of failure of the considered structure (CUR 141, 1990)

increas-The probabilistic approach allows designers: (i) take into account the uncertainties of the input parameters and treat them as random variables; (ii) describe the failure of the structure in various possible failure modes based on the physical processes of these fail-ure mechanisms; and (iii) find a probability of failure of the whole system taking account all individual elements (cross-sections and/or structures) and just if it is acceptable; For instance, for a flood defence system, the accepted probability of flooding is not the same for protected region It depends on the nature of the protected area, the expected loss in case of failure and the safety standards of the country For this reason accepted risk is a better measure than an accepted failure probability because risk is a function of the probability and the consequences The failure probability and the probability of the con-sequences form the probability part of the risk When the risk is calculated the design can be evaluated For this, criteria must be available such as a maximum acceptable probability of a number of casualties or the demand of minimizing the total costs includ-ing the risk For determining the acceptable risk one needs to refer to a frame of refer-ence This frame of reference can be the national safety level aggregating all the activi-ties in the country After the evaluation of the risk one can decide to adjust the design or

to accept it with the remaining risk Thus, the probabilistic approach is an integral sign method for the whole system

de-However, the number of applications of probabilistic design and risk analysis are still ited because of its complexity in use This study reviews existing knowledge of probabil-istic design and risk based approach; develops tools and models that are applicable to the case of a coastal flood system Methods and proposed models can be further devel-oped/ modified to apply widely for other types of engineering structure system More re-searches and applications of these design methods in civil engineering design are there-fore relevant Especially in developing countries such as Vietnam, China, India, etc where the modernization and urbanization is rapid, all new design works should be under careful consideration in the sense of reliability and in terms of optimization Optimiza-tion can be done by balancing the benefit with initial investment, repair and mainte-nance costs and consequences of potential failures Final choices should take into account also the developing characteristics i.e limited initial investment, fast economic growth and cheap labour availability The probabilistic design approach is capable to solve these above issues

lim-1.6 Fundamentals of this study

1.6.1 Problem definition

In low-lying coastal regions, such as in the Netherlands, Japan, Vietnam, and desh, the coastal flood defence systems (CFDS) are very important to protect the low-lying regions from sea flood Commonly, this system contains sea dikes, dunes, estuarine river dikes, dams, sluices, pumping stations, etc (see Figure 1.1) In principle, failure of any component of the system leads to flooding of the protected area (also referred to as

Bangla-a polder)

high ground

city low lying

Figure 1.1: Figure 1.1 Example of a coastal flood defence system (after Vrijling 2001)

For any coastal flood defence system, there often exist three following questions:

Question 1: What is the actual safety of hinterlands/polders and do we know if the

sys-tem is reliable enough?

Question 2: How safe is safe enough? Do we need to improve the safety level (level of

protection) and to which level of protection is the “best choice” (e.g 1/100, 1/250, 1/500,

1/1000 or 1/10,000 per year)?

Question 3: Given a pre-defined level of protection, what would be the “best design” for

the system (and of course, best design for the system components, too)?

1.6.2 Study objectives and approach

Efforts of this study are to establish a generic probabilistic framework to answer three questions above and to give emphasis on specific applications of the framework to a case study in Vietnam

The first question can be answered by implementing a reliability safety assessment of the considered system This can be done by assessing of all system elements which considers

all possible failure mechanisms of these elements by comparing the strength of the ments and joint loads acting on it The safety level of the whole system can be easily de-termined by a combination of failure probabilities of these considered system elements

ele-This will be called in this study as “reliability safety assessment model” (RSAM)

The second question can be addressed by doing a risk assessment and evaluation of the whole system For various protection levels the cost of protection and the consequence of system failures need to be quantified In general the first term is increased when the level

of protection is higher while the second term is inversed It is possible to search for the point which provides the minimum total of these two components This leads to a con-

cept of “risk-based optimization model” (RiOM) The result from this risk analysis

will be used as an input condition for answering the next question

Having a certain (or pre-defined) safety level of protection, various geometric alternatives for the coastal flood defences can be generated In order to answer the last question, for each of those the failure probability and the total cost of construction need be deter-mined The failure probability of each alternative can be determined based on the previ-ous reliability safety model The total cost of construction here includes initial invest-ment cost and the maintenance-repair cost during the total service lifetime of the struc-tures The investment cost can be estimated based on a given geometry while the later term can be expressed through the failure probability of the each considered alternative Again the minimum point of total construction cost in space of failure probability is

looked for This algorithm and procedure is, in this study, called “reliability-based

op-timization model”(ReOM)

Important input for these previous models is the load boundary condition For coastal

flood defences the loads are mainly defined by hydraulic loads (water levels, wave tions) These loads are often not certain values and will be quantified by using statistical analysis of observation data or on the basic of physical means of coastal processes The combination of these two methods can lead to a joint hydraulic boundary condition The strength of structures can be determined based on design geometry and materials used Other important issues are investigating of failure mechanisms of the CFDS and quantifying the consequences of the failure

condi-In addition, other important aspects which serve in the safety assessment model and

reliability base optimization model (RSAM and ReOM ) are limit state equations

(LSF) of failure mechanisms The LSF is established on basis of its physical process- based equation In order to come up with a reliable and appropriate result an overview of all possible failure mechanisms and their physical processes is necessary More insight into dominant/main failure mechanisms would be appreciated and can be achieved by reviewing literature, using analytical & numerical methods and testing with physical ex-periments This study focuses on the first two approaches The later is outside of scope

of this study

Combining all these above study components leads to an overall probabilistic safety, ability and risk based approach for CFDS, which aims at answering the research ques-tions The overall study approach is conceptually illustrated in Figure 1.2

reli-1.7 Scientific and social relevance

Optimal CFDS

Reliability based design model:

optimal geometry given [Pf]

{from component to system level}

Risk-based design model:

optimal level of protection [Pf]

Component/system reliability analysis;

Safety assessment (P f ; P f sys )

Reliability based design model:

optimal geometry given [Pf]

{from component to system level}

Risk-based design model:

optimal level of protection [Pf]

Component/system reliability analysis;

Safety assessment (P f ; P f sys )

Failure modes

Limit State Eq.

Optimal CFDS

Reliability based design model:

optimal geometry given [Pf]

{from component to system level}

Risk-based design model:

optimal level of protection [Pf]

{system level}

Reliability based design model:

optimal geometry given [Pf]

{from component to system level}

Risk-based design model:

optimal level of protection [Pf]

Component/system reliability analysis;

Safety assessment (P f ; P f sys )

Component/system reliability analysis;

Safety assessment (P f ; P f sys )

Failure modes

Limit State Eq.

RSAM

Figure 1.2 Overview of study models

1.7.1 Scientific relevance

The scientific contribution of this study can be mainly identified as:

- an extension of data management techniques and regional frequency analysis methods in extreme value analysis of coastal data Findings have been published

in for example van Gelder and Mai Van (2008), van Gelder et al (2008), Mai Van

- investigation of present situation of coastal flood defences and coastal protections

in the Red River delta of Vietnam (Mai Van et al 2009f, Mai Van & Pilarczyk 2005);

- developing a generic probabilistic and risk-based design model for flood defence and widening its applications to a case of coastal flooding in Vietnam, see for in-

stance Mai Van et al (2009b,c,g,h); Mai Van et al (2008a,c), Mai Van et al (2006a,c)

Proposed approaches in this study can be modified and further improved to apply to various problems of civil engineering systems and issues of engineering systems

Parts of this thesis have been presented at over ten scientific international conferences and post published in their proceedings

For a whole list of the related scientific papers which have been published while doing this research please see Appendix 1

1.7.2 Social relevance

The research results contribute considerably to sea dike improvement and rehabilitation works which are on-going in Vietnam These methods and findings can be used as sup-porting tools in decision making processes in domains of flood defences and flood safety, e.g to build up safety standards and regulations, selection of relevant level of flood pro-tections, which are necessary in many low-lying regions

This study gives also considerable input to the Delft Cluster project (in the Netherlands) and the FLOODSite project (under Framework 7, European Commission)

The research presented in this thesis is focused on developing and widening the tion of probabilistic design, safety and reliability assessment, risk assessment and evalua-tion and risk based design in the fields of flood defences An application is done for the case of a coastal flood defence system in Vietnam On basis of the study objectives, the thesis includes 9 chapters, as below A schematic outline of this thesis is presented in Figure 1.3

applica-Chapter 2 gives an overview of the flood defence and sea flood defence history of nam A description of the actual coastal protection strategy, present situation of the coastal defence systems in the Red River delta is made Recognition of problems and proposal of research items and a preliminary proposed coastal protections strategy are presented in this chapter

Viet-Methods and application of statistical techniques to analyse extreme environmental data are presented in Chapter 3 Methods concern data management techniques (i.e trend analysis, stationarity tests, seasonality analysis and peak over threshold) and statistical techniques to model the occurrence probabilities of extreme values, including e.g Re-gional Frequency Analysis and Mixed Distribution Fitting An application is made to de-termine the maximum sea loads i.e extreme waves and maximum storm surges levels at the locations of interest for the case study in Vietnam

In the subsequent chapter, Chapter 4, the theoretical backgrounds of probabilistic design and reliability analysis are critically reviewed and investigated Safety assessment and re-

1.8 Structure of this thesis

liability based design models are developed for specific application to the coastal flood defence systems Length effects to system reliability are discussed as well

In Chapter 5, application of the theoretical methods that were given in Chapter 4 is made for the case study of Nam Dinh coastal flood defence system, Red River delta of Vietnam The safety of the existing system is investigated The weakest link of the sys-tem and dominant failure modes for Nam Dinh sea defence system are found A set of optimal geometry dimensions for the dike system are presented in accordance with analysis results from the Reliability based design model

An extensive overview of risk analysis methods, risk evaluations and assessments, basic risk related concepts and their developments are discussed in Chapter 6 Methods and applications to establish acceptable risks and safety criteria are reviewed The actual framework of acceptable risk and risk based optimal safety for flood defences worldwide and in the Netherlands is investigated Possible applications of the methods to the Viet-namese situation are discussed

Chapter 7 and Chapter 8 present applications of the theory which are given in Chapter 6

to the Vietnamese situation In Chapter 7, a risk evaluation is made and risk criteria are set for Vietnam by means of individual and societal risks National acceptable risk is es-tablished basing on available data of Vietnam The risk criteria on national scale are transformed to apply to the Nam Dinh polder in evaluating flood risk and proposing safety standard for the Nam Dinh coastal region The optimal safety is re-checked and confirmed by application of an economic risk based approach in Chapter 8

Conclusions and recommendations of the research are presented in Chapter 9 This cludes remarks regarding the methodology: strong and weak points of present research range of applicability and cautions; as well as remarks about the application case of Viet-nam

in-Probabilistic design of coastal flood defences in Vietnam

Conclusions

Sea boundary conditions

Introduction

Chapter 1 Introduction, background and

fundamentals Chapter 2 Water defences in Vietnam & Description of case study

Chapter 3 Establishing hydraulic load conditions of coastal area Application of Nam Dinh coast

Theory review and discussion

Chapter 4 Reliability analysis and reliability based

design

Chapter 6 Risk analysis and risk based design of

water defence system

Chapter 9 Conclusions and recommendations

Figure 1.3 Schematic outline of the thesis

Chapter 2

This chapter aims at describing the current situation of the coastal zone in the Red River Delta of Vietnam and assessing its present protective measures The coastal zone

in the Red River Delta of Vietnam is under large threat due to fluvial flooding, coastal flooding and coastal erosion In the region, sea dikes have been used as the predominant counter measure with two intended functions, viz protecting low-lying areas from coastal flooding; and reducing risk for the hinterland caused by erosion However, the sea dikes seem not to function well and are insufficient to withstand the hydraulic design condi-tions of sufficiently low frequency of occurrence To have a better insight into the actual situation of the whole system, this chapter, firstly, investigates the historical development

of the coastlines in the delta based on available information Secondly, hydrodynamic and morphological processes of the coastal and estuarine systems are reviewed and fur-ther analysed Subsequently, an effectiveness assessment of the present protection works

is presented Finally, further research needed to improve the rehabilitation of coastlines and the safety of the regions is discussed on the basis of the actual analysis results Analysis result of this chapter is important input for further researches which find out a coastal protection strategy optimally to reduce the vulnerability of the coastal region

DESIGN PRACTICE

Vietnam, situated in the tropical monsoon area of the South East Asia, is a prone and flood vulnerable country A large part of the population is mainly involved in agricultural and fishery sectors which are situated in the low lying river flood plains, del-tas and coastal margins In addition, important ports are located along the coast On the other hand these areas are the most important potential disaster areas due to flood-ing and land erosion (Wijdeven,2002) Typhoons from the South China Sea bring tor-rential rainfall and high winds to the coast and further inland On average four to six ty-phoons attack the coast annually The monsoon season coincides with the typhoon sea-son resulting in heavy damage, loss of life, and destruction of infrastructure facilities and services

typhoon-The reason that flood disasters are so serious lies in the fact that most of the population lives in areas susceptible to flooding, either originating from the rivers or from the sea The main population centres and intensively cultivated lands in the Red River and Me-kong deltas and the narrow connecting coastal strip of the country are particularly vul-nerable to flooding from monsoon rains and typhoon storms In addition, due to the dy-namical nature of coastal and river processes land erosion is occurring seriously at many places Thus fluvial and coastal flooding and associated land erosion are the most impor-tant potential disasters facing Vietnam (Pilarczyk and Nguyen,2002)

Because of limited budgets and unavailable access to technology (proper expertise and technical facilities), in the past shore management and flood control were limited to sim-ple and cheap approaches such as manual labour, training in dike inspection, procedures for evacuating the population during emergencies, strategic positioning of supplies for emergency repairs, providing relief after a disaster and accepting land retreat

As a result of the severe sea loads and the rather low safety level of the present dikes, the flood defence system of Vietnam fails regularly Since 1953, Viet Nam was affected by numbers of flood disasters, each disaster responsible for the loss of hundreds of lives and considerable damage to infrastructure, crops, rice paddy, fishing boats and trawlers, houses, schools, hospitals, etc… The total material damage of the flood disasters over last 60 years exceeded $US 7.5 billion Additionally, floods and storms caused the loss of more than 20,000 lives (ADRC, 2006 & DDMFC, 2007) The most severe storms and floods induced disasters occurred in North Vietnam in 1971, 1996 and 2005; in the South

in 1997; and in the Central in 1964 and 1998 Mostly, these events were initiated by phoons which attacked the coastal zones then, additionally, accompanied by heavy mon-soon rains inland

ty-The relatively low safety level of the sea dikes in Vietnam was noticed in 1996 during two visits of Dutch expertise missions (DWW/RWS, 1996a,b) Most designs of the sea dikes in Vietnam are based on loads with return period 20 year or even shorter periods Compared to the Dutch standard (return periods 1000 to 10000 year) these return peri-ods are very small Besides this fact the Dutch mission marked that most Vietnamese dikes were designs as poor and disputable As a result the true probability of failure of

the Vietnamese flood defence system exceeds by far the design frequency (Mai et al.,

2006c, 2007d) Although designed to fail once in 20 years the sea defense system might well fail almost every year The experiences in the past 20 years support this statement Besides of these above imperfections in the designs, it should be noted that the adopted return periods are not based on proper statistical analysis Often adopted return period

20 years is founded on a rather arbitrary basis However, these arbitrary considerations already show a notion of the fact that the safety level of important, valuable areas should be increased compared to the safety level of less important areas This system re-flects logical results, which could have been obtained by common risk analysis Future improvements of flood safety standards might build on the existence of this system However, these improvements should be based on proper risk analysis of the areas under consideration

An additional important fact is the economic situation of Vietnam The country is idly developing with limits resources for improvement of the flood defence system On the other hand, this situation asks for a more detailed and careful analysis to ensure that

rap-2.2 Coastal protection in the Red River Delta: current strategies

the limited resources are used in the optimal way which takes into account the ing characteristics (limited initial investment, fast economic growth, and cheap labour) Currently, Vietnam is seeking to build a modern system of shoreline protection, dike re-habilitation and flood risk management based on new design methodologies and tech-nologies, such as using more advanced design and construction methods (often adopted from foreign countries), in combination with state-of-the-art warning and forecasting sys-tems to improve the emergency preparedness, and by the construction of storm-proof shelters for disaster impact reduction These objectives are incorporated into the recently initiated Vietnam flood defence development plan, where safety against flooding and sus-tainable land protection is recognized as fundamental requirements for all forms of devel-opment There is great emphasis placed on strengthening and improving the entire sys-tem of sea dikes to prevent sea flood damages The importance of dike safety has long been recognized as fundamental for the continued development of Vietnam

develop-The Red River delta is known as the most important region for all economic activities in the north of Vietnam, where the majority of the region's population is concentrated and supports nearly half the country's rice production, is under threat due to fluvial, coastal flood and coastal erosion In the delta there are six river mouths distributing over ap-proximately nearly 200 km along its coastlines, which leads to a complex coastal estua-

rine system Due to the impacts of natural processes from the sea (waves, storm surges,

currents, sediment transport, and sea level rise) the coastlines have experienced dynamic

changes and occasionally destruction due to episodic events (typhoons) At present sea

dikes are used as the predominant coastal protection measures with two intended tions: (i) protecting low-lying areas from coastal flooding; and (ii) reducing risk for the hinterland caused by erosion However, the sea dikes seem not to function well and are insufficient to withstand the hydraulic design conditions of sufficiently low frequency of occurrence Consequently, the risk of damage and land loss is still high Recently, the ty-phoon Damrey, which occurred on 27th September 2005 and is considered as the most vigorous in the last 50 years, causing sea dike breaches, sea-water flooding and serious loss of land The need of finding a more proper coastal protection strategy to reduce risk for the whole region is now being expressed by both local and national responsible au-thorities in Vietnam

STRATEGIES

2.2.1 Description of study area

The Red River Delta is characterized as low lying with an extensive network of river branches and with long stretches of dikes and sea defences (Figure 2.1) There are six ac-tive river mouths namely Thai Binh, Ninh Co, Tra Ly, Ba Lat, Van Uc, and Day The

Ba Lat river is the main branch of the Red River discharging its water into the Gulf of Tonkin, South China Sea

Figure 2.1 Position of the Red River Delta in Vietnam

Most of the coastal regions of the Red River delta have an elevation of less than one ter above mean sea level This fact makes the coastal area vulnerable to sea flooding and salt water intrusion The deltaic coastal areas and low-lying coastal strips to a distance

me-of about 20 km landwards are threatened because me-of the combined occurrence me-of storms, surges from the sea and a high flow discharge from the rivers Along the coastline, sea dikes are the most dominant coastal structures protecting the behind regions from sea flooding

Sea dikes are built over the centuries mainly through local initiative The sea dikes are generally poorly designed and poorly constructed Due to the bad state of the dikes a significant part of the yearly budget has to be allocated to repairs and maintenance Generally, two main and interrelated problems are distinguished in project areas: (i) seri-ous erosion of the coastlines and (ii) heavy damage to the coastal flood defences by storms As stated in the “Annual reports on losses and damages of coastal regions” from

1976 to 2005 by the Department of Dike Management and Flood Control (DDMFC), more specific problems are as follows

Severe structural longshore erosion and foreshore erosion takes place along the tected coastlines and protected coastlines respectively This structural erosion is due to a lack of fluvial sediment supply to the shore necessary to compensate for the net natural sediment longshore transport occurring during the North-East monsoon while foreshore erosion is due to longshore and cross-shore sediment transport which takes place during storms The structural erosion rate is about 10 to 20 metres per year while the foreshore erosion leads to a deepening of the foreshore in front of the sea dikes ranging from 0.3 to

unpro-2.2 Coastal protection in the Red River Delta: current strategies

0.6 metres per year This may lead to rapid retreat of the shoreline if appropriate counter measures are not undertaken in-time

Beach erosion, dike breaching due to typhoons, storm surges, and wave action caused treat of up to 3000m of the shoreline during the last 100 years The total area of land loss is approximately 18,000 ha (nearly as large as the current area of the Hai Hau dis-trict)

re-Typhoons often attack coastal regions with wind-strengths of 9 to 12 Beaufort causing house collapse, loss of lives and property In the last 30 years, from 1976 to 2005, storms demolished about 7000 houses and a hundred fishing ships Additionally, a thousand of people died

Due to dike breaches, seawater has over flown the hinterland resulting in flooding of and salt intrusion in cultivated land Statistics shows that nearly 40,000 ha cultivated land was impacted by salt, and over 100,000 tons of food was lost Salt mining fields, and shrimp hatching ponds were also heavily damaged

Storm surges often accompanied with high tides caused damage of Nam Dinh sea dikes almost every year During the period from 1976 to 2005 about 1.900.000 cubic metres of earth and 1.000.000 cubic metres of stone were taken away from the sea dikes Therefore the expenditure on maintenance is very large (in the order of millions of Euros) Heavy damage and collapses of the defensive system, especially the dike system and re-vetments, has occurred Many sections of dikes and revetments failed and were breached induced by a variety of failure modes This caused flooding in a wide area along the coastlines and as a consequence, it led to loss of land, economic values and even to loss

of human life

It appears that coastal erosion and damage of the defensive system lead to many effects

on the social and economic development in the area In response the central and local authorities have undertaken some efforts in order to restrain the possible adverse conse-quences Also, some sections of new sea dikes have been built However, due to budget constraints and a lack of a suitable design methodology, such efforts still remain limited

to reactive and temporary measures instead of implementing structural and long-term lutions

so-Figure 2.2 shows the recent situation of sea dike failures at Nam Dinh province due to typhoon Damrey Figure 2.3 provides evidence of fast coastline retreat due to heavy ero-sion at Hai Trieu-Hai Ly communities, in Hai Hau district, Nam Dinh province in the Red River delta As can be seen from Figure 3 the normal village life was there in 1995, however just 6 years later the village was totally removed due to erosion

Figure 2.2 Serious damage and breach of sea dikes in Hai Phong province after Saola typhoon in

2005

Figure 2.3 Hai Trieu Village in 1995 and it was disappeared in 2001

2.2.2 Causes of coastal erosion

The most dynamic part of the coastal zone in the Red River delta lies in the centre of the Nam Dinh province The coastal zone of Nam Dinh is roughly 80.000 hectares in size and has about 70 kilometres of coastline This coastline is naturally divided into 3 sec-tions by 4 large estuaries: the Ba Lat (Red River main reach), Ha Lan (So River – has been cut-off some ten years ago), Lach Giang (Ninh Co River) and Day (Day River) From the North to the South these are: Giao Thuy section: from Ba Lat estuary to So estuary belonging to Giao Thuy district, about 27 kilometres; Hai Hau section: from So estuary to Ninh Co estuary, belonging to Hai Hau district, 27 kilometres; Nghia Hung section: from Ninh Co estuary to Day estuary belonging to Nghia Hung district, 16 km long (see Figure 2.4)