Luận án tiến sĩ: Probabilistic design of Coastal flood defences in Vietnam

Probabilistic Designof

Coastal Flood Defences in Vietnam

Probabilistic Designof

Coastal Flood Defences in Vietnam

MAI VAN CONG

Master of Science in Hydraulic Engineering

geboren te Ninh Bình, Vietnam.

Prof drs ir LIK VrijlingCopromotor:

Dr it PALAJ.M van Gelder

Samenstelling promotieconnnissie:

Rector Magnificus, voorzitter

Prof drs ir JAK Vijling ‘Technical University of Delft, promotorDr ir P.H.A.JLM van Gelder Technical University of Delft, copromotorProf dr ir MILF Stive Technical University of Delf

Prof dr H Hanson Lund University, SwedenProf dr J.W Hall sweastle University, UKIr K.W Pilarcayk Formerly RijkswaterstaatDr ir $.N Jonkman “Technical University of Delft

Prof dr W.S.J Uijttewaal ‘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 975-00-9025688-1

Printed by: Sieea Repro, Delft, the Netherlands

Cover layout: Pham Quang Diew

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

to my family

Probabilistic design of coastal flood defences in Vietnam.

‘The year 2005 brought hurricane Katrina to New Orleans (U.S), it was also a historicalyear 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 inInuman casualties anel enormons economic damage In response to that, both the centraland 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 latestavailable knowledge at all levels Vietnam has a profound practical experience in the fieldof flood protection; however, there is a lack of up-to-date knowledge in the field of dikedesign and flood risk management, especially regarding to coastal flood defences Thisstudy focuses on coastal flood defences in Vietnam and its objectives are to evalnate thecurrent level of safety and reliability, to develop safety standards and design practicesand 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 firstlyreviewing, developing and widening the application of probabilistic design and state ofthe art of tisk assesement 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 Vietazun whieh 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 coastaÏ ares ng approaches for flood dolene‘Vietnam and a description of the case study area ‘The coastal zone in the Red RiverDelta of Vietuam, where the case study was catried out, is under a potential threat ofuvial flooding, coastal flooding and coastal erosion, In the region, sea dikes have beenused as a predominant countermeasure with two intended functions, Le, 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, thesca dikes do not seem to fimetion very well and their strength is insufficient to withstandthe hydraulic londs with a sufficiently low frequency of occurrence, This is due to theunderestimation of the hydranlic boundary condlitions and imperfections in the design,construction and maintenance of the dike systems A qualitative multi-criteria analysishas been developed for proposing a suitable coastal protection strategy for the studyarea, It is proposed to use a combination of sea dike systems to provide flood protection,

Summaryand beach erosion control measures, stich as groins and/or sand nourishments to preservea shallow foreshore,

To have a better understanding of the actual hydraulic load conditions in the coastalarea and to derive input information for the reliability analysis, methods for the statisti-cai analysis of extreme hydraulic load conditions have been developed and applied

(Chapter 3) These include data management techniques (ie 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 therandom environmental variable with very small probability p) Hydraulic observationsfor the coast Vietnam are available for a period of 35 years and on that basis an estimateneeds to be derived for a design sea load, ie, 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 (MDP) are used in this study for sueh a sea load,In onder to ensure that the statistically predicted values of the interested sea loads willbe as close as possible to the true values, recommendations are made to compare thesevalues with physical hindeast 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 mouths, 12 months and 224 months in the

data The 6-month periodicity reflects well the influence of two monsoons per year to theMuctuation of sea water level: the 12-month periodicity is explained by the influence ofyearly typhoons in the region The 224-month periodicity, which is approximately 18.6

years, is almost exactly equal to a tidal citele,

Although sea level rise (SLR) is not a focus of this study, based on a detected lineartrend of the observed water level data the rate of SLR can be estimated approximatelyat 0.68 em per year at Nam Dinh coast In addition, this study recommends that estima-tion of SLR should be based on not ouly mathematical extrapolations, but also the loug-term nature of sea level fluctuation,

‘The theoretical background of probabilistic design and reliability analysis of flood fences is reviewed in Chapter 4 Models for safety assessment and reliability based designare developed for the specific application to coastal flood defence systems In the safetyassessment mnodel, present probabilistic calculation techniques are applied to quantify thefailure probability of the system and the system components The effect of the length ofthe 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 geometey of the dike system

de-given an admissible failure probability.

Chapter 5 demonstrates the application of the models proposed in the previous chapterto Vietnam The ease study area is the Nam Dinh coastal flood defence system in theRed River delta of Vietnam, It is found that the existing coastal flood defeuce system isless safe than its required safety standards According to the stancards the dikes shouldbbe designed to provide a target safety of 1/20 per year However, the total failure prob-ability of the dike system is estimated to be much higher, about 0.15 per year This corresponds 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,

which likely contributes about 44% to the total failure of the dike system Instability ofannour units ofthe rovetinents gives an influence of 259% Overllowing and geotechnicalinstability of inner slope have smaller, but still a considerable contribution (10 and 20%,respectively) Piping, geotechnical instability of the outer slope and toe foot instabilityhave a negligible influence on the safety A sensitivity analysis shows that wave heightand water level are the most important loading parameters The findings are inline withthe practical experiences and observations from dike failures in recent years that oc-curred in the Nan Dink dike system.

‘The application of the reliability based design model shows that if the existing safetystandard (1/20 per year) would be applied, the geometry of the current dikes would beInsufficient For instance, the crest level of the dike required from the reliability baseddesign is 6.80 meters, whereas the existing crest level of the dike is 5.25 to 5.50 metersAn extensive overview of approaches for risk assessment and risk evaluation is given inChapter 6 In general, risk is defined as the product of probability of occurrence and con=sequence of disasters Lo 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 theacceptance of risk by the publie and the decision makers Presently, the acceptable flooding risk is judged from two points of view which relate to the acceptance of immdation‘conseqnences The first is acceptable risk regarding loss of life, which is under judgementof 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 theVietnamese situation In Chapter 7, anapproach for estimation of the level of acceptableflood risk for Vietnam is propased The approach is based on earlier framework of Vi-‘ling (2000) in which acceptable risk level is based on characteristics of the activity(benefit, voluntariness) and baseline mortality statisties in a countey 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 therisks 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 nationallevel of acceptable risk of Vietnam is established, Evaluation of the present floods on thenational seale shows that acceptable societal risk of Vietnam is five to ten times higherthan in the Netherlands,

To obtain a risk limit for the Nam Dinh coastal region, the nationally acceptable risk basto be distributed over all the regions in one country A societal risk limit for Nam Dinhis 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

"Individual risk: The probability (per year) of being killed at certain location assuming pertnent presence of the population,

' Societal risk: The probability of exceedance (per year) of an accident with a certain number offatalities Societal risk is often shown by means of an FN curve It displays the probability peryear of accidents with N of more fatalities,

SummaryDinh is protected to a safety standard with exeeodanee probabilities lower than 1/150per year.

‘The optimal safety is evaluated based on an economic approach for risk based design inChapter 8 Historical data on economic loss due to floods in Vietham is available since1970 The distribution of these losses can best be approximated by a lognormal distribution and the yearly expected value of the damage equals approximately SUS 589410” peryear This value is comparable with the actual flood damage of Vietnam during the last

‘years (over the the last 10 vears the total flood damages on average are about 1.0% of

the GDP of Vietnam per year)

Economic risk analysis of the Nam Dinh dike ring shows that the safety standard of 1/50-yoars should be applied to the present economic situation (Situation 1) Because of thefast 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 Dink as well as in Vietnam appears to be able according to some existing risk limits that have been referred to literature Theseresults 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 risksare acceptable or aalditional risk reducing measures are necessary ‘The methods and re-sults presented in this thesis provide an important input background for making thesedecisions.

unnecept-Delft, 2010 July 29

Mai Van Cong

Probabi tich 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 eatastrofale impact van ty:foons op de kustverdediging en kustgebieden In totaal hebben acht tyfoous de Vietua-ese kust bereikt, Deze resulteerden in menselijke slachtoffers en enorme econormischeschade Als repsous op deze gebeurtenissen hebben de lokale en centrale overheden vanVietnam zeer veel aanclacht besteed aan het herstel van de waterkeringen langs de kusten de verbetering van de veligheid van de kustgebieden

‘Voor de verbetering van het waterkeringyystoem van Vietnam is de meest recente kennis‘op verschillende niveaus benodigd Vietnam heeft uitgebreide praktiikervating op het ge-bied van hoogwaterbescherming Eehter, er is cen gebrek aan up-to-date kennis overdijkontwerp en de bebeenjng van overstromingsrisico's, specifiek voor kustgebieden Dezestudie richt zich op đe waterkeringen in de kustgebieden van Vietnam, De doelen betref"fen het evalueren van het huidige veiligheidsniveau en de betrouwbaatheid van de water"keringen, het ontwikkelen van veiligheidsuormen en ontwerprichtlijuen en het formulerenan cen voorstel voor de verbetering van de waterkeringen.

‘Om deze doelen te berciken heeft het onderzock zich ten eerste gericht op het evalueren,“ontwikkelen en verbreden van de toopassing van probabilistisch ontwerpen en risicoanaly:se voor waterkeringen (het theorie deel) Vervolgens is, mede in het kader van kennisover-dracht, de voorgesteldo annpak toegepast op het waterkeringssysteem Iangs de kust vande provineie Nam Dinh in Vietnam Dit gebied is representatief voor het kustsysteen in‘Vietnatn (het toepassings dee).

De belangeifkste onderwerpen die in dit proefschrift zijn behandeld zijn hieronder

Hoofdstuk 2 geeft een overzicht van het kustgebied, de bestaande wijze van bescherming in Vietnam en eon overzieht van het ease studie gebied De kustzone in dedelta van de Rode Rivier in Vietnam, waarvoor de ease studie is nitgevoerd, wordt be-drvigd door rivieroverstromingen, kustoversteoningen en lesterosie, In het gebied hebbenwaterkeringen langs cle kust twee funeties, namelijk het beschermen van Inaggelegen ge-bieden tegen overstroming en het beperken van de risieo’s voor het achterland door kus.terosie Eehter, tijdens recente tyfoons is gebleken dat de waterkeringen Tangs de kustlet goed funetioneren, De sterkte van de keringen is zelfs onvoldoende om hydrauliscebelastingen 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 beleer en onderhouel van dijksystemen Eien kwvalitatiovemulti-criteria analyse is ontwikkeld om geschikte strategieén voor kustverdediging vanhot gebied te bepalen Er wordt voorgesteld om een combinatie van zeedijken voor be-

hoogwater-* This “Samenvatting” is translated by Bas Jonkman and Wim Kanning (TUDelt from its PhụcTish version,

Summaryscherming tegen averstromingen, en kusterasie beperkende maatregelen, zonls strandhhoolen en zandsuppleties toe te passen om 20 cơn fiauwe vooroever the handhaven.ˆMethoden voor de statistische analyse van extreme hydraulische belastingen in het kust

agcbied zijn ontwikkeld en toegepast (hoofdstuk 3) Hiermee ontstaat een beter begrip van

do daadwerkelijke hydranlisehe belastingen en komt informatie beschikbaar waarmee eendetrouwbaarkeidsanalyse kan worden uitgevoerd De beschouwde methoden betreffen+eehnieken voor data beheer (trend analyse, stationariteits testen, seizoenscyclus analyse‘en peaks-over-threshold analyse) en statistische methoden om de kans op het voorkomenvan een extreme waarde (de waatde die wordt overschreden door de willekeurige omge-vvingavariabele met een zeer kleine kans) te bepalen, Hydraulische observaties voor đe-kust van Vietnam zijn beschikbaar over een periode van 35 jaar Op basis daarvan dienteen schatting gemaakt te worden van cơn ontwerp belasting, bestaande uit golven enstormopzet, die overcen komt met een herhalingstijd van 100 tot 10.000 jaar De metho-den Regionale Frequentie Analyse (RFA) en het fitten met Gemengde Verdelingen zijngebruikt voor hot afleidden van de hydraulisehe belastingen voor het kustgebied, Om zekerte stellen dat de statistisch voorspelde waarden van de relevante belastingen dicht bij dedaadwerkelifke waanden liggen, zijn aanbevelingen gedaan om deze warden te vergelj-ken met đe resultaten van fysische modelevaluaties van historische gebeurtenissen Dezezijn gebaseerd op historische kenmerken van tyfoons, zoals de windsnelhetd.

Hoewel de analyse van zeespiegelstijging niet de focus van deze studie betreft,blijkt datde zeespiegelstijging lineair verloopt Op basis van de beschikbare gegevens is de absolutee0spiogcletjging ingeschat op 0,68em per jaar voor de kust van Nam Dinh Daarnnast isaanbevolen om de inschatting van zeespiegelstijging niet alleen te baseren op wiskundigeextrapolatie van gegevens, maar ook op de langjarige cyeli die het zeespiegelniveau bepa-Ten,

De thearetische achtergrond van het probabilistise ontwerp van waterkeringen is schouvd in hoofilstuk 4 Modellen voor đe veiligheidstoetsing en op betrowwbaarheid e-baseorl ontwerp zijn outwikkeld voor de specificke toopassing voor waterkeringssystemenTangs đe kust In het model voor veiligheidstoetsing zijn probabilistische rekentechnieken

be gebruikt om de kans op het falen van het system en systeemcomponenten te

kwantifice-ren Het effect van de lengte van het dijkensysteem op de betrouwbaarheid van het systeem is in het model ook meegenomen, Het model voor op betrouwbaarheid gebaseerdontwerp is gebruikt om een optimale geometrie van het dijksysteem af te leiden witgnan-de van een toelaatbare faalkans,

Hoofdstuk 5 demonstreert de toepassing voor Vietnam van de modellen die in het vorigehhooflstuk zijn behandeld, Dit is gedaan in een ease studie voor de Rode rivier delta inde provineie Nam Dink in Vietnam Het blifkt dat het waterkeringsysteem langs do kustminder veilig is dan wordt beoogel met de bestaande ontwerpnormen Volgens deze nor:‘men zouden de dijken ontworpen moeten worden om een veiligheid van 1/20 per jaar tebieden, 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 cen 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 system Instabilitet van steenbekledingen drangt 25% bij loop en geotechnische instabiliteït van het binnentalud hebben een kleinere, maar nog

Over-steeds aanzienlijke bijdrage aan de faalkans (10%, respectievelifk 20%) Piping en

geo-technische instabiliteit van het buitenland talud en de dijkteen hebben een bare bijdrage Uit een gevoeligheidsanalyse blijkt dat de golfhoogte en de waterstand debelangtijkste belasting variabelen zijn.

verwaneloos-De bovindingen liggen in lin met praktijkervaringen en observaties betreffencle recentedijkdoorbraken in het dijkensysteem van Nam Dinh De toepassing van het model voorop betronwbaarheid gebaseerd ontwerp laat zien dat de afmetingen van de bestaandedijken onvoldoende zijn om te voldoen aan de huidige velligheidsnorm (1/20 per jaa)Het blijkt bijvoorboeld wit de botrouwbnarheidsgebaseerd ontwerp dat de benodigde krinhoogte 6.80 meter moet zijn, terwijl de bestaande kruinhoogte 5.25 tot 5.50 meter be-range.

Fen nitgebreid overzicht van methoden voor risicoanalyse en risico evaluatie is gegeven inhoofdstuk 6 Over het algemeen wordt rsico gedefinieerd als het product van de kans ophet optreden van rampen en de gevolgen daarvan In deze studio is het oerstroiningerisi-

co beschouwd in termen van slachtofferrisico (individueel” en maatschappelijk rsieo") en

ceconomische schade (economisch risico) Een centrale vraag in risico management betreftde aanvaardbaatheid van risico’s voor het publiek en beleidsmakers De aanvaardbaas-hheid van overstromingstisico’s wordt over het algemeen vanuit twee gezichtspunten be-ordeeld Het eerste betreft de aanvaardbaarheid van slachtofferrsieo’s, wel voor indivviduen als grote groepen Het tweede geziehtspunt heeft betrekkking op een economischebalans van kosten en baten.

In hoofdstuk 7 en 8 is do in hooklstuk 6 gepresenteerde theorie toegepast op Vietnam Inhnooflstuk 7 is een aanpak voor de bepaling van het aanvaardbare niveau van overstro.tmingstisico’s voor Vietnam voorgesteld De aanpal is gebasoerl op eon eerder ranmnwol:van Vrijling (2000) waarin het acceptabele risico niveau wordt bepaald op basis van dekenmerken van de activiteit (baten, vrijwilligheid) en statistioken voor sterfte door diver-se oorzaken in een Inne Hierbij wordt allereerst de sterfte door natunlijke oorzaken enreguliore activiteiten, zoals deelname aan het verkeer, besehouved Op basis van beschik-bare statistioken, is de basis sterftekans voor Vietnam bepaald inelusief de risico’s diesamenhangen met het verkeer en het leven in overstromingsgevoclige gebieden in Viet~Een FN curve voor overstromingen in Vietnam is afgeleid en deze toont de kans op ge-betrteniseen met bijbehorende nantallen slachtoffers, Het raamvverk voor risico evaluatievan Vaijling (2000) is aangepast en het nationale acceptabele tisiconiveau voor Vietnamis afgeleid, Evaluatie van het overstromingstisico op nationale schaal laat zien dat hetacceptabele maatschappelifk risiconivean voor Vietnam vijf tot tien maal hoger is dan‘voor Nederland.

‘Om een tisiconorm vast te stellen voor het kustgebied van Nam Dinh, is het nationaalaanvaardbate rsico verdeeld over alle regio's ia het land, Hiermee wordt een risiconorm

ndividueelsiieor de kans (per jae) om om het leven te Komen op een zekere locate titgaaaddevan permanente nanwezigheid van de bevolking,

" Maatschappeliirsico: de overscijdingskaus (per jar) van een oageval met meerdere offers Mnatschappelijkrisco wordt vaak algebeeld in cơn FN curve Deze toont de kans per jaar

slacht-‘op ongevallen met N of meer slachtoffes

Summary‘voor Nam Dinh verkregen en het blijkt dat het huidige overstromingsrisico ver boven denorm ligt Volgens de norm is het overstromingstisico aaavaardbaar als langgelegen kust-gebieden een veiigheidsnorm krijgen van ten minste 1/150 per jaar

Het optimale veiligheidsniveau is geévalucerd in hoofdstuk 8 op basis van een sche benadering voor risico-gebascerd ontwerpen Historische gegevens over economischeschade door overstromingen in Vietnam zijn beschikbaar sinds 1970 De statistisce verdoling van deze schades kan het best benaderd worden met cen lognormale vordcling De

economi-larakteristieke waarde van de jaarlijkse schade is ongeveer geljk aan SUS 89x10" per

jaar Deze waarde is vergelijkbnar met de daadwerkelijk opgetreden sehade door overstromingen in Vietnam in recente jaren Gedurende de laatste tien jaren bedroegen deschades door overstroming gemiddeld 1,5% van het bruto nationaal produet văn Viet-‘Een economische riscoanalyse (optimalisati) voor het đkensyateem van Nam Dịnh latzien dat cen veiligheidsnorm van 1/50 per jaar gekozen zou moeten worden uitgaandevan de huidige economische situatie Rekening houdend met sulle economisle ontwikke-lừng in de toekomst wordt een veligheidsnorm van 1/100 per jaar voorgesteld

Het maatschappelifk risieo door overstromingen in Nam Dinh en de rest van Vietnamlijkt niet aanvaardbaar volgens risiconormen die zijn afgeleid op basis van de Titeratinr.Deze vesultaten wijzen op de noodzaak van meer maatschappelifke discussie over en ver=der onderzock naar het aanvaardbare nivean van overstromingstisic’s in Vietnam Be-sluitvorming is nodig over de aanvaardbaasheid van de bestaaurde risico’s en de noodzaakvan aanvallende maatregelen De methoden en resultaten die zijn gepresonteerd in dit

proefschriftleveren belangrijke achtergrond informatie aan om deze besluiten te nemen,

Delft, 29 jnli 2010

Mai Van Cong,

CHAPTER 1 ISTRODUCTION

LL General hackgromd on food defences

12 Developments in the design and safety aseasment of Nood defanes.

18 Catalysts in now policy towards flood risk

LT Soval relevance"`.

(CHAPTER 2 COASTAL MANAGEMENT AND FLOOD DEFENCES IN VIETNAM24 Plood defences iu Vi i design practice

22 Coastal protection in the Red River Delta: current strategies

221 Description of study aca

222 Causes of coastal erosion

2.2.3 Historia! information on coastline changes

2.24 The year 2005 of typhoons and its consequences

28 lyrdtynamis in the gulf of Tonkin,

24 Longhone sediment transport and coastline changes.

25 Actual salty ofthe coastal protections26 Dewsim

27 Conclusions

Contents(CHAPTER 8 ESTABLISHING THE COASTAL HYDRAULIC CONDITIONS.

41 Introduction

32 Rogôoual fequeney analysis

83 Data management methods i estreme vale analysis

33.4 The Annual Maxima (Gumbel) Method

44.2 Peak-overtveshold in modelling extreme value data - °3.39 Trend analysis, sttionarty tes end detection of seasonality and natural fluctuation34 Statistical background use in REA.

B41 Definitions, steps im REA and site discordance detection

84.2 L-Moment statistics

84.3 Discordance measure

B44 Robust distances

35 REA of extreme sa water Joel long the Vietnamese East sa const

45.1 Etmme sea water level data

35.2 Regional quency analysis ofthe sen date

85.3 Diseussion onthe oppiation of RFA.

56 Statistical model of extreme sea water lve at Nam Din const.

871 Conceptual model and steps8.7.2 Example of implementation38 Discusion.

(CHAPTER 4 RELLABILITY-BASED DESIGN: THEORY,AA Iusraducton

42 Definition af an engineering system

$3.1 Non incar Z-fnetion and design values4.32 Non normally distrsbuted basic variables4⁄4 — Analysis of a aston

$41 Basic systems

45 Reliability of a systenn,

46 Langth effets nnd system reliability

46.1 ypothetca ease: a đắc system comprises ne-type eros section

os

421 Application model 1: Reliability safety assesment “

4:72 Appliation model 2: Retiability based optimal design 100

4:73 On the probabilistic models ond reliability eneulasion methods tới(CHAPTER 5 RELIABILITY-BASED DESIGN OF COASTAL PLOOD DEFENCES IN VIBTNAM 109

5l Introduction 100

5:2 Description of Nam Dink coastal foe defences, nó

Pane fatluse mechan of he sen dike syste 12` 14

55 Rely analysis of the Nam Dinh se dike system nh

BSL Safety assessment ug

2 Paul tre analysis and system faire probability - "5",

I Sensitivity analysis of stochastic variables 1gSection length effects 14

‘Sytem length effets 19

Total faitre probability of Hai Hove đc system 131“Transtation of sytem safety to desi frequency of chat sections, 192

54.8 Toll failure probability of Nam Dink coastal flo defence system 134

6.95 Estimation of potential loss of bf due to fle 159

6.3.6 Eemomi optimisation 16064 Dhewsinm 182

64.1 Risk related to los of ie 102

"

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

TA Less of life due to flooding in Vietnam

718.1 Individual risk for coastal flooding in Vietnam132 Sovetl risk đục to flood ơn ä country level

TA Caco study: analysis of flood risk in Nam Dinh eonstal zoneTAA Loss of hfe estimates for Nam Dink coastal zone

14.2 Acceptable fod rst for Nam Dinh coastal zone143 Discussion

(CHAPTER 8 APPLICATION OF ECONOMIC RISK-BASED DESIGN 1 Eeonomieleses of Vietnam dục to footing

82 Bomdary conditions,

S21 Cost estimatesB22 Sea wuter levels

5.2.3 Dike height in relation to exoedance frequency of water lve

83 Economic optimisation of Nam Dinh sea ond defenesBIE Cost optimal function

8.9.2 General asrrnptions

8.8.3 Determination of potential economic risk

S95 Sensitivity analysts

SA Extension of the van Dante model5.4.1 Basic nhe: van Dantig model (1956)S42 Protected orca wath inland slope543 Sensitivity analysis

wor208

9.13 Iovestgation ond development of safety aseasment and reliability based design

2.14 Risk based design with a special opplication to the Vietnamese situation9.1.5 Appliation tothe Vietnamese constal flood defences.

92 Reconmndations

9.2.1 Reliability ond safety assessments922 Acceptable risk

2.23 Establishment of sea boundary conditions.

924 The Nam Dinh sea defences and optima safety of Nam Dink coastal zone825 Coastal yrotetion strategies forthe Red River delta of Vietnam

2.2.6 Vietnam load risk and flood defence strates

REFERENCES

LIST OF FIGURES Lisr oF Tames.

239243245

Chapter 1

1.1 GENERAL BACKGROUND ON FLOOD DEI ENCES

‘Coastal and fluvial areas bring enormous benefits for human life, however both areas areliable to potential risk the sea and rivers, Le, storm waves and surges, tsunamis, highriver discharges and consequently floods, To avoid flooding risks while living in these ar-fas our ancestors adapted to nature recognisably In coastal arvas huunan settlements‘were in the higher natural areas or on man mace 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 sit on theland, 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 Iying areas were takenInto use At the same time, provisions were put into place in relation to agriculture todrain the land; wells were made for drinking water and peat was dug up i many placesfor fuel and salt production, The consequence was a fallin the level of the ground surface, as a result of which flooding became a greater problem,

In addition, a rise in population is often accompanied by a rise in huanan actives in theflood-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

In response, dikes were constructed Initially, they only supplemented the naturalhhoights; in the rivers area for instance, perpendicular to the bank walls to redirect riverwater flowing outside the banks along populated areas to lower-lying back lands Iu theconstal area the influence of the sea increasod steadily due to sea level rise and Iandl sub-sidence, which means that, more and more lowlands had to be protected by dikes (Ton-neijek & Weijers 2008)

Generally, almost all low-lying coastal and/or fluvial land around the world is denselypopulated 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,

l5

Dut also the level of required protection inereases The value of property logically creases and the nunber of huanan lives threatened by ñooding increases as well The con-sequences of a flood will, therefore, be much more serious than before as the imundationdepth and the protected values of the area increased during the centuries

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

low-1.2 DEVELOPMENTS IN THE DES!

FLOOD DEFENCES

SN AND SAFETY ASSE: SMENT OF

In the past, one of the first solutions to overcome flooding problems was to live with ter There was no protection against floods coming from the sea, the lakes or the rivers.In ease of a high water level people simply moved up if necessary and stayed on highergrounds where possible.

wa-In the beginning of the Middle Ages, more proactive action was taken in order to copewith flooding At fist small high ground were constructed and offered the population arelatively safe place People and their livestock could survive a period of flood while theirland around would get covered with a fresh and fertile layer of clay.

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

‘The first dikes were designed based on experience and skill ofthe builder The experienceand skill was gathered in practice Practice was to build the dike at the highest knownwater 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 bysurges during storms

In the last century, mathematical and statistical knowledge improved Combined withthe introduction in practice of fluid and soil mechanies the approach for dike design be-came more and more seienti e hydraulic load on a flood defence could be predicted‘more accurate and the strength of the flood defences could be calculated, always basedon extreme value extrapolation of the existing situation.

In the Netherlands, after the disaster in 1953 a statistical approach to the storm surgelevels was chosen and an extrapolated storm surge level formed the basis for dike design.It was also deseribed how the Delta Committee optimised these safety levels, which wereexpressed in terms of the return period of high water levels which must be withstood bythe primary dike system (van Dantzig, 1956).

16

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 mechanisuns of a dike ora 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 Basternscheldt in 1976 and later in thedesign of a storm surge barrier in the Nieuwe Waterweg,

In approximately 1979 project was started to apply the probabilistic methods to thedesign of dikes in general (Vrijling, 2001) The development of 2 complete approach towater defense systems took a considerable time (Tonneijek & Weijers, 2008) Recently.the approach was tested applied on many Dutch polders or dike rings The probability ofNooding of these polders was caleulated and insight in the weak spots was gained Re-sults can be viewed in detail e.g, in Veijling et al, (2001), Jonkman et al, (2008) andFLORIS reports’

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 grownconsiderably 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 notonly occurring in the Netherlands, but also in many other flood prone countries Gener-ally, the number of discussions on how sale is safe enough has been of increasing atten-tion nowadays worldwide

Probabilistic design with risk based design concepts are considered the most modern

ap-proaches in the field of flood defences Advantages of the methods are that allows

de-signer to take into aecount the uncertainties of the input parameters as random 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 failture modes can be considered These all help to determine the true probability of flooding,of a protected area and to judge its acceptability in a view of the consequences of theprotected area Thus, probabilistic appronch is apparently an essential tool to analythe flood defence systems.

vari-Although the probabilistic approach has been increasingly proposed and applied in thelast years in the West (see e.g the concept, method and application in Bakker & Vrijling1980; Viiling et al 1908; Hall et al 2000; Oumeraei et al 2001 and Vortman 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 al2007b) Thus, widening the applications of the method to the Vietnamese situation offlood defences and coastal protection is therefore necessary The methods can also be ap-

plied to cases of other developing countries.

* FLORIS: Floodrisk and Safety assessment of primary flood defences in the Netherlands1

1.3 CATALYSTS IN NEW POLICY TOWARDS FLOOD RISK

One of the tasks of human civilisation is to protect individual members, groups of peopleand 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 occurrenceof the hazard which had shown the consequences The modern probabilistic approachaims to give protection when the risks are assessed to be high, Risk is defined as theprobability of a disaster e.g a flood related to the consequences As long as the modernapproach is not firmly embedded in society, the idea of acceptable risk may be quitesuddenly influenced by a single spectacular accident or ineident.

For example in the Dutch case, after the sea flood dlisaster happened in 1953, the Dutchsociety got a severe wake up cal This event showed the vulnerability of the country andthe lack of science to base dike design on, After this event the iauplementation of sophis-ticated techniques was accelerated The Delta Committee (a committee that advised onthe strategy needed to prevent disasters like the one that occurred in 1953), proposed anew 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 thescientific approach ofthe flood defences One of the important changes was the develop-‘ment of probabilistic techniques in hydraulic engineering, Using statistical tecuniques forthe 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, hmzvicane Katrina, whieh hit New Orleans, induced new policy and attentionof the U.S to the sea defences, Moreover, the influence of Katrina on flood risk manage-‘ment poliey is not only valid for the U.S, but also for tany other flood prone countries.In Netherlands, the New Orleans food was critically reviewed to understand what hadhappened (Kok et al., 2007) Lessons learnt partially stimulated the establishment of anew Dutch Delta Committee.

Ta Europe, more attentions and efforts are paid to reducing flood risk Tinplementation ofthe BU floods Diective is under way Govemment taskforees are preparing emergency as

‘wells longterm management for lood An example would be the FLOODSite project’

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

= preliminary flood risk assessment

- the preparation of flood risk maps

~ the preparation (and implementation) of lod risk management plans.

Preferences for other types of measures ie insurance, evacuation, using compartmentdikes and room for water ate also under discussion in Europe.

FLOODSite: EU funded project under the Framework 7 (2004-2009)18

1.4 Additional stimulus of this research

1.4 ADDITIONAL STIMULUS OF THIS RESEARCH

In Vietnam, the year 2005 was considered a historical year because of the disastrous inpact of typhoons on its coastal regions In total eight typhoons hit the Vietnamese coastsin that year resulting in human easwalties and large economic damage Especially, ty-phon Damrey was of exeeptional strength and was one of the heaviest typhoons of thelast 3 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 economie loss of over 50 Million USD (DMWG, 2005) As a wake:

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

Improvement ofthe flood defence system in Vietnam calls for the use of present availableknowledge on all levels Vietnam has profound practical experience inthe fields of ood

protection; however, there is lack of recent theoretical knowledge in the fields of eke de

sign, safety and reliability, risk-based approaches in flood defences ‘Therefore, the for of these wplodate knowledge and experiences was strongly recommended(DWW/RWS, 1996; Vaijling ef al, 2000: Mai et a, 2006g) to Vietnam.

rans-In an initial attempt to improve safety for the coastal low-lying rogions and rehabilitatethe sea dike systems on the long rum, a huge sea dike research program has been initiatedby 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 programare an assessinent of the present situation of the sea defence regarding the safety and theestablishment of new safety standards and developing a master plan for protecting thelow-lying coastal regions The theoretical background and technical knowledge in design-ng of sea flood defences and establishing sea houndary 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 Viet~namese situation in flood defences, in general, and sea flood defences, in particular

1.5 PROBABILISTIC VS DI ERMINISTIC APPROACH.

‘The conventional design method in civil engineering design is a deterministic approach,‘The basis of the doterministie approach is the so-called design values for the loads andthe strength parameters Loads for instance are the design water level and the designsignificant wave height, Using design rules according to codes and standards it is possibleto determine the dimensional geometrical parameters of the civil engineering structuresi.e buidings, bridges, tunnels, dams, dikes and storm surge barriers, These design rulesare, 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 marginbetween the design value of the load and the characteristic value of the strength is large

19

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

sÿs-Probabilistic design with reliability: and risk- based design concepts have been increas:ingly proposed and applied in the fields of civil engineering and flood defences during theInst years A fundamental difference with the deterministic approach is that the probabikistic design methods are based on an acceptable frequency or probability of failure of the

considered structure (CUR 141, 1990).

‘The probabilistic approach allows designers: (i) take into account the uncertainties of theInput parameters and treat them as random variables; (ii) describe the failure of thestructure in various possible failure modes based on the physical processes of these failure mechanisins; and (ii) ind a probability of failure of he whole system taking accountall individual elements (cross-sections and /or structures) and just if it is acceptable; Forinstance, fora flood defence system, the accepted probability of flooding is not the sameor protected rogion It depends on the unture of the protected area, the expected loss incease of failure and the safety standards of the country For this reason accepted risk is aDetter 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

cou-sequences form the probability part of the risk When the risk is calculated the designcan be evaluated, For this, criteria noust be available such as a maximum acceptableprobability of a number of casualties or the demand of minimizing the total costs ineluc-ing the risk For determining the acceptable risk one needs to refer to a fraine of refer-cence This frame of reference can be the national safety level aggregating all the actties in the country After the evaluation of the risk one can decide to adjust the design orto accept it with the remaining risk Thus, the probabilistie approach is an integral de-sign method for the whole system,

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 tothe 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, ete.where the modernization and urbanization is rapid, all new design works should be tudercareful 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 accountalso the developing characteristies Le, limited initial investinent, fast economic growth‘and cheap labour availability The probabilistic design approach is eapable to solve theseabove issues.

lim-0

1.6 Findamentals of this study

1.6 FUNDAMENTALS OF THIS STUDY

1.6.1 Problem definition

In low-lying coastal regions, such as in the Netherlands, Japan, Vietnam, and dosh, 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, estuarineriver dikes, dams, sluices, pumping stations, ete (see Figure 1.1) In principle, failure ofany component of the system leads to flooding of the protected area (also referred to asa polder)

Figure 1: Figure 1.1 Example of a coastal ood defence system (after Vrijling 2001)

For any coastal flood defence system, there often exist three following questions:westion : What is the actual safety of hinterlands /'polders and do we know ifthe sựs-‘tem is reliable enough?

Question 2 How safe is sale enough? Do we need to iinprove the safety level (level ofprotection) 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 2: Given a pre-defined level of protection, what would be the “best design” forthe 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 easestudy in Vietnam.

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

2L

all possible failure mechanismms of these elements by comparing the strength of the ‘ments and joint loads acting on it The safety level of the whole system can beeasily de-termined by « combination of failure probabilities of these consideted system elements.‘This will be called in this study as “reliability safety assessment model” (RSAM).‘The socond question can be addressed by doing a risk assessment and evaluation of thewhole system For varions protection levels the cost of protection and the consequence ofsystem fatlures need to be quantified In general the first term is increased when the levelof protection is higher while the second term is inversed Tt is possible to search for the

ele-point which provides the minimnm total of these two components, This leads to a

con-copt of “risk-based optimization model” (RiOM) The result from this risk analysiswill be nsed as an input condition for answering the next question,

Having a certain (or pre-defined) safety level of protection, vasious geometric alternativesfor the coastal flood defences ean be generated In order to answer the last question, forcach of those the failure probability and the total cost of construction need be deter:mined The failure probability of each alternative ean be de ed based on the provi-‘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-‘ures, 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 islooked for This algoritlun 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 coastalood defences the loads ate mainly defined by hydraulic loads (water levels, wave conditions) These loads are often not certain values and will be quantified by using statisticalanalysis of observation data or on the basie of physical means of coastal processes Thecombination of these two methods can lend to a joint hydraulic houndary condition,‘The strength of structures can be determined based on design geometry and materialsused Other important issues are investigating of failure mechanisms of the CFDS and

‘quantifying the consequences of the failure.

In addition, other important aspects which serve in the safety assessment model andreliability base optimization model (RSAM and ReOM ) are limit state equations(LSF) of failure mechanisms The LSF is established on basis of its physical processbased equation In order to come up with a reliable and appropriate result an overview of,all possible failure mechanismns and their physieal processes is necessary More insightinto dominant/main failure mechanisms would be appreciated and can be achieved byreviewing literature, using analytical ác numerical methods and testing with physical ex-

periments This study focuses on the first to appronches The Inter is outside of scope

of this study,

‘Combining all these above study components leads to an overall probabilistic safety, reliability 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

2

nai geometry men TP optnal level of protection (Pf)

[trom componct ta sydem rel {tan level)

| Optimal CFDS:

igure 1.2 Overview of study models

1.7 SCIENTIFIC AND SOCIAL RELEVANCE,

17⁄1 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 publishedin for example van Gelder and Mai Van (2008), van Gelder ot al (2008), Mai Van.et al, (2007a),

= an extension of van Dantzig’s economic risk-based model to enable its applicationor sloping polders

~ am extension of existing probabilistic reliability models for flood defence systemsxegarding length effets of the gystem (see also Mai Van & Calle 2009);

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

in the Red River delta of Vietuam (Mai Van et al 2009f, Mai Van & Pilarcayk.

= 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,¢.g,h); Mai Van et al, (2008a,e), Mai Van et al,

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

1.8 STRUCTURE OF THIS THESIS

‘The research presented in this thesis is focused on developing and widening the tion of probabilistic design, safety and reliability assessment, risk assessment and ovalua-tion and risk based design in the fields of flood defences An application is done for thecease of a coastal flood defence system in Vietnam On basis of the study objectives, thethesis includes 9 chapters, as below A schematic outline of this thesis is presented inFigure L3

appliea-Chapter 2 gives an overview of the flood defence and sea flood defence history of Viet.nam A description of the actual coastal protection strategy, present situation of theconstal defence systems in the Red River delta is made Recoguition of problems andproposal of research items and a preliminary proposed coastal protections strategy are

presented in this chapter.

‘Methods and application of statistical techniques to analyse extreme environmental dataare presented in Chapter 3 Methods concern data management techniques (i.e trendanalysis, stationarity tests, seasonality analysis and peak over threshold) and statisticaltechniques to model the oceurrence probabilities of extreme values, including eg Re-

sional Frequency Analysis and Mixed Distribution Fitting An application is made to

de-termine the maxim sea Toads ie extreme waves and maximum storm surges levels atthe locations of interest for the case study in Vietnam,

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

a

1.8 Structure of this thesis

liability based design models are developed for specific application to the coastal flooddefence systems Length eflects to system reliability are discussed as wel

In Chapter 5, application of the theoretical methods that were given in Chapter 4 ismade for the ease 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-‘tein and dominant failure modes for Nam Dinh sea defence system are found A set ofoptimal geometry dimensions for the dike system are presented in accordance withanalysis results from the Reliability based design model.

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

situation are discussed,

Chapter 7 and Chapter 8 present applications of the theory whieh are given in Chapter 6to the Vietnamese situation In Chapter 7, a risk evaluation is made and risk eriteria aresel for Vietnam by means of individual and societal risks, National acceptable risk is es-tablished basing on available data of Vietnam The tisk eriteria on national scale aretransformed to apply to the Nam Dinh polder in evaluating flood tisk and proposingsafety standard for the Nam Dinh constal region The optimal safety is re-checked andconfirmed 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‘age of applicability and cautions; as well as remarks about the application ease of Viet-

in-Probabilistic design of coastal food defences in Vietnam.

igure 1.3 Schematic outline ofthe thesis

26

Chapter 2

COASTAL MANAGEMENT AND FLOOD DEFENCES IN VIETNAM

This chapter aims at describing the current situation of the coastal zone in the RedRiver Delta of Vietnam and assessing its present protective measures The coastal zonein the Red River Delta of Vietnam is under large threat due to fluvial flooding, coastalooding and coastal erosion Iu the region, sea dikes have been used as the predominantcounter measure with two intended functions, viz protecting low-lying areas from coastalfooding: and reducing risk for the hinterland caused by erosion However, the sea dikesseem not to funetion well and are insufficient to withstand the hydraulic design condi-tions of sufficiently low frequency of occurrence To have a better insight into the actualsituation of the whole system, this chapter, firstly, investigates the historieal developmentfof the coastlines in the delta based on available information Secondly, hydrodynamicand morphological processes of the coastal and estuarine systems are reviewed and fur-‘ther analysed Subsequently, an effectiveness assessment of the present protection worksis presented Finally, further research ueeded to improve the rehabilitation of coastlinesand 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 acoastal protection strategy optimally to reduce the vulnerability of the coastal region.

2.1 FLOOD DEFENCES IN VIETNAM: CURRENT SITUATION AND

DESIGN PRACTICE

‘Vietnam, situated in the tropical monsoon area of the South Bast Asia, is a typhoonprone and flood vulnerable countey A large part of the population is mainly involved in

agricultural and fishery sectors which are situated in the low Iying river flood plains,

del-tas and coastal margins, In addition, important ports are located along the coast Onthe other hand these areas are the most important potential disaster areas due to fooding and land erosion (Wijdeven, 2002) ‘Typhoons from the South China Sea bring torrontial rainfall and high winds to the coast and farther inland On average four to six tý»

phioons attack the coast annually The monsoon season coincides with the typhoon

sea-son resulting in heavy damage, los of life, and destruction of infrastructure facilites and

‘The reason that flood disasters are so serious lies in the fact that most of the populationlives 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 val-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 tangplaces Thus fluvial and coastal flooding and associated land erasion are the most impor-‘tant potential disasters facing Vietuam (Pilarczyk and Nguyen, 2002).

Because of limited budgets and unavailable access to technology (proper expertise andtechnical facilities) in the past shore management and flood control were limited to im:ple and cheap approaches such as manual labour, training in dike inspection, proceduresfor 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 londs and the rather low safety level ofthe present dikes, theflood defence system of Vietnam fails regularly Since 1953, Viet Nam was affected bynumbers of flood disasters, each disaster responsible for the loss of hundreds of lives ancconsiderable damage to infrastructure, crops, riee paddy, fishing boats and trawlers,houses, schools, hospitals, ete The total material damage of the flood disasters overlst 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 andfloods induced disasters occurred in North Vietnam in 1971, 1996 and 2005; in the Southin 1997; and in the Central in 1964 and 1998 Mostly, these events were initiated by ty»

phoons which attacked the coastal zones then, additionally, accompanied by heavy mon:

soon rains inland.

‘The relatively low safety level of the sea dikes in Vietnam was noticed in 1996 duringtwo visits of Dutch expertise missious (DWW/RWS, 1996a,b)- Most designs ofthe seadikes in Vietnam are based on londs with return period 20 vear or even shorter periods.‘Compared to the Dutch standard (return periods 1000 to 10000 year) these return perieods are very small Besides this fact the Dutch mission marked that most Vietnamesedikes 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 frequeney (Mai ef al,

2006e, 20074) 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 adoptedreturn periods are not based on proper statistical analysis Often adopted return period20 years is founded on a rather arbitrary basis, However, these arbitrary considerationsalready show a notion of the fact that the safety level of important, valuable areasshould be increased compared to the safety level of less important areas This system re-Aeets logical results, whieh could have been obtained by common risk analysis, Futureimprovements of flood safety standards might build on the existence of this system,However, these improvements should be based on proper risk analysis of the ateas underconsideration

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-28

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 develop,ing characteristics (limited initial investment, fast economic growth, and cheap labou),‘Currently, Vietnam is seeking to build a modern system of shoreline protection, dike re-hhabilitation and flood risk management based on new design methodologies and tech-nologies, such as using more advanced design and construction methods (often adoptedfrom 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-proofshelters for disaster impact reduction These objectives are incorporated into the recentlyinitinted 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-tein of sea dikes to prevent sea Mood damages, The importance of dike safety has longbeon recoguized as fundamental for the coutinued development of Vietnam,

‘The Red River delta is known as the most important region forall economic activities inthe north of Vietnam, where the majority of the region's population is concentrated andsupports nearly half the country’s rice production, is uneler threat due to fluvial, coastalflood and coastal erosion In the delta there ate six river mouths distributing over ap-proximately nearly 200 kn 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 seadikes are ased as the predominant coastal protection measures with two intended fune-tions: (i) protecting low-lying areas from coastal flooding; and (ii) reducing risk for thehinterland caused by erosion However, the sea dikes soem not to function well and areinsufficient to withstand the hydraulic design conditions of sufficiently low frequency ofoccurrence Consequently, the risk of damage and land loss is still high Recently, the ty-

phòn Daunrey, which occurred on 27° September 2005 and is cousidered as the most

Vigorous in the last 50 years, causing sea dike breaches, sea-water flooding and seriousloss of land The need of finding a more proper coastal protection strategy to reduce riskfor the whole region is now being expressed by both local and national responsible au-thorities in Vietnam.

2.2 COASTAL PROTECTION IN THE RED RIVER DELTA: CURRENT

2.2.1 Description of study area

‘The Red River Delta is characterized as low lying with an extensive network of riverbranches and with long stretches of dikes and sea defences (Figure 2.1) There are six ae-tive river mouths namely Thai Bình, Ninh Co, Tra Ly, Ba Lat, Van Ue, and Day TheBa Lat river is the main branch of the Red River discharging its water into the Gulf ofTonkin, South China Sea,

)

Figure 2.1 Position of the Red River Delia in Vietham

‘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 andsalt water intrusion The deltaic coastal areas and low-lying coastal strips to a distanceof about 20 kin landwards are threatened because of the combined occurrence of storms,surges from the sea and a high flow discharge from the rivers Along the coastline, seadikes are the most dominant coastal structures protecting the behind regions from seaooding.

me-Sea dikes are built over the centuries mainly through local initiative, The sea dikes aregenerally poorly designed and poorly constructed Due to the bad state of the dikes asignificant part of the yearly budget has to be allocated to repairs and maintenanceGenerally, two main and interrelated problems are distinguished in project areas: (i) serifous erosion of the coastlines and (ii) heavy damage to the coastal flood defences bystorins As stated in the “Annual reports on losses and damages of coastal regions” from1976 to 2005 by the Department of Dike Management and Flood Control (DDMEC),‘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 erasion is due to anck of fluvial sediment supply to the shore necessary to compensate for the net naturalsediment longshore transport occurring during the North-East mousoon while foreshoreerosion is due to longshore and cross-shore sediment transport which takes place duringstorms, The structural erosion rate is about 10 to 20 metzes per year while the foreshoreerosion leads to a deepening of the foreshore in front of the sea dikes ranging from 0.3 to

unpro-30

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 appropriatecounter 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 landloss is approximately 18,000 ha (nearly as large as the current area of the Hai Haw dis-tet).

re-‘Typhoons often attack coastal regions with wind-strengths of 9 to 12 Beanfort causinghouse collapse, loss of lives and property In the last 30 years, from 1976 to 2005, stormsdemolished 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 Nooding of aadsalt intrusion in cultivated land Statisties shows that nearly 40,000 ha cultivated land‘was impacted by salt, and over 100,000 tous of food was lost Salt mining fields, andshrimp hatching ponds were also heavily damaged.

Storm surges often accompanied with high tides caused damage of Nam Dinh sea dikesalmost every year, During the period from 1976 to 2005 about 1.900.000 cbic metres of‘earth and 1,000,000 cubie 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 revvotments, has occurred Many sections of dikes and revetments failed and were breachedinduced by a variety of failure modes This caused flooding in a wide area along thecoastlines and as a consequence, it led to loss of land, economic values and even to lossof human life,

It appears that coastal erosion and damage of the defensive system lead to many effectscon the social and economic development int the area In response the central and localauthorities 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 budgetconstraints and a lack of a suitable design methodology, such efforts still remain limitedto reactive and temporary measures instead of implementing, structural and Jonge term sơ-Tutions.

Figure 2.2 shows the rocent situation of sea dike failures at Nam Dinh province due totyphoon Damrey Figure 2.3 provides evidence of fast coastline retreat due to heavy ero-sion at Hai Trien-Hai Lycommunities, in Hai Haw district, Nam Dinh province in theRed River delta, As can be seen from Figure 3 the normal village life was there in 1993,however just 6 years later the village was totally removed due to erosion.

aL

Figure 2.5 Hai Trieu Village in 1905 and it was disappeared in 2001

2.2.2 Causes of coastal erosion‘The most dynamic part

the Nam Dinh provinee

and has about 70 kilometres of coastline This coastline is naturally divided into 3 seetious by 4 large estuaries: the Ba Lat (Red River main reach), Ha Lan (So River ~ has‘been cutoff some ten years ago), Lach Giang (Ninh Co River) and Day (Day River)From the North to the South these ate: Giao Thuy section: from Ba Lat estuary to Soestuary belonging to Gino Thuy district, about 27 kilometres; Hai Hau section: from Soestuary to Ninh Co estuary, belonging to Hai Hau district, 27 kilometres; Nghĩa Hungsection: from Ninh Co estwary to Day estuary belonging to Nghia Hug district, 16 kmlong (see Figure 2.4).

f the coastal zone in the Red River delta lies in the centre ofhe coastal zone of Nam Dink is rou 30.000 hectares in size

a2

2.2 Coastal protection in the Red River Delta: current strategies

HANAM PROVINCE:NINH BINH PROVINCE

Figure 2.4 Nam Dinh coastlines and its current situation [Soutee: PDD Nam Dĩnh, 2003]

In this section the causes of erosion for the Red River Delta coastal area are discussed.In Figure 2.4 we observe two accretion areas, one north and one south of the eroded areaof Hai Han district Tt should be noted that accreted areas usually are stabilized by local

people reducing the natural distribution of sediment along the coast This could have

ac-celerated the erosion rate in the eroded areas of Hau Hau In eroded areas the beach israther narrow with approximately 100 - 200 metres at low tides According to the records of the Provincial Dike Department (PDD) of Nam Dinh province, the total aver-

aged retreat distance during the last 60 years (from 1952) at the Hai Hau coasts is

esti-‘mated at 1100 metres,

‘The main reasou for the heavy erosion along Hai Hau coastline is due to changes in thedistribution of the natural fluvial sediment discharge to the whole system In the end ofthe 19° century the Ngo Dung River (presently named So River) lost its importance as amajor branch of the Red River This could have been a result of either a natural devel-‘opment or due to manmade river training works, which changed the Red River flow di-rection, The Ngo Dung River was dammed in 1955 because of the low flow discharge inthis branch, These caused the total loss of sediment transport to the beaches of Han Hart

Another reason is expected to be due to the construction of the Hoa Binh hydropowerđam upstream of the main input branch (Da river) of the RRD (1979-1994), The mainobjectives of the project are electricity generation, which currently contributes nearly

20% of the national electricity production of Vietnam (in 2000) The dam project,

thereby, regulates about 70% flow discharge in the Red River system (inereasedl 38

down-stream river discharge in the dry season and decreased risk of flooding: downdown-stream inthe rainy seasou) However, this has reduced about 50% of the total natural suspendedsediment concentration of the downstream river system at Son Tay station as shown inFigure 2.5 (Son Tay station is located at about 100 kilometers downstream of Hoa Binh,Dam, it measures water levels, river discharges and suspended sediment concentration ofthe Red River), This led to a reduetion of half the sediment supply to the coastal areas,In summary, the construction of the Hoa Binh dam could have been an accelerating fnc-tor for coastline erosion along the RRD's coastal areas

In addition, most sediment deposited in the mouth of Red River is stabilized by local

people (inpoldering for new agricultural areas) with consequences that less sediment can

be distributed along the coast,

2.2.8 Historical information on coastline changes

‘On a smaller scale, historical information on coastline changes and coastal protection in‘Nam Dinh is available since 1890, The available information is in forms of reports andravings, which includes the major events of coastline retreat due to erosion and dikebreaches In the period of 1890 to 1972, there were no measurement data,

‘More extensive information is available since 1972 Observations of cross-shore profiles of‘Nam Dinh dikes were made at several locations enabling an analysis of the situation in‘more detail Since 1972 beach profile measurements took place once every 1 to years.‘During the period from 1890 to 1971, due to erosion and erosion induced dike weakening,olten dikes breached after severe typhoons As a consequence the coastline shifted inland

”

2.2 Coastal protection in the Red River Delta: current strategies

about 850 meters Six occurrences of dike breach and associated constline retreat are ported during those 80 years.

re-During a period of 30 years from 1972 to 2008, event related constline retreats were

ap-proximately 400 meters on average Six dike breaches occurred and three reconstructions

of the dikes were undertaken These can he observed clearly from the overlapping shore profiles in Figure 26.

crose-Figure 2.6 Overlapped cross shore profiles from 1972 to 2003 at Hai Ly-Hai Tieu section

Averaged retreat rate of the coastline during the period of 30 years is shown in Figure2.7 The retreats of foreland based levels of -0.50, 0.0, +0.50 (+MSL) are indicated Thebase point is the position of the foreland level 0.0 in the year 1972 Based ou the averagetrend from the above analysis, and if no proper measures for mitigation of erosion areundertaken, the situation would continue to be similar to the trend of the last period.‘This means that after every 10 years the retreat of coastline would be around 150 metersinland According to the current trend, every 10 years the location of the dikes wall haveto shift 200 motors landward.

— Beebe EE BER

Figure 2.7 Averaged constline retreat distance ofHai Han coast based on observations from 1972ta 20 by Dike Department of Nam Dinh province

2.2.4 The year 2005 of typhoons and its consequences

‘The year 2005 was a historical year because of the disastrous impact of typhoons on seadefences in Vietnam In total 8 typhoons hit the Vietnamese coast in this year resultingin human casualties and large economic damage Typhoons No 2 and No 7 were of exceptional strength and belong to the heaviest typhoons in the last 5 decades,

Typhoon No.2, named Saola, made landfall on 31 July 2005, with wind forces nearingthe storm eye at Beaufort seale 9 (75-88 kan per hour) hitting mainly the coastal areas 0FQuang Ninh and Hai Phong provinces, resulting in several kilometres of sen dike dam-ages, especially on the island Cat Hai, where 8 km dikes were heavily damaged and bro.ken and which needed completely rehabilitation,

‘Typhoon No 7, named Damrey, is considered as the most vigorous in the last 50 yearsWith wind force in the storm eye at Beaufort seale 12 (118 to 133 km per hour) and the‘wind gusting above Beaufort Scale 12, the typhoon attacked all coastal provinces of theRed River delta of Vietnam By 5.00 AM local time on 27 September the wind force rosefrom Beaufort scale 7 to scale 12 (133 kilometres per hour) hitting provinees in the area1s the typhoon moved in a west to northwest direction, ‘The typhoon caused high stormsurges, which coincided with high tide.

‘The wave run up was as high as 3-4 metres High storm surge in combination with hightide led to extensive overtopping of sen water at sea dikes in almost all affected prowinees, It broke certain sea dike sections in Hai Han district in Nam Dinh province andHau Loe district in Thanh Hoa province Total damage was enormons while 28 kilome-‘res of sea dikes were seriously broken In Nam Dinh a stretch of 800 metres sea dikeswas completely washed out (Figure 2.8).

‘The seawater penetrating inland by 3-4 kilometres in coastal provinces and the followingfash floods in upland areas destroyed at least 1,194 houses and damaged another 11.576.‘More than 130,000 ha of rice fields have been submerged and damaged, most of whichcould not been harvested before the typhoon Severe damage occurred to transport andcleetricity infrastructure and particularly to irrigation systems According to the rapidassessment of the Disaster Management Working Group the direct damage estimate as of

28 September 2005 is estimated at SUS 500 million (UNDP, 2005).

36

2.8 Hydrodynamies in the gulf of Tonkin

2.3 HYDRODYNAMICS IN THE GULF OF TONKIN

‘The Gulf of Tonkin is connected to the South China Sea in the Pacific Ocean It fences a tropical climate with monsoon winds It is a semi-enclosed sea with an averagewater depth of 50 meters During the winter monsoon the wind blows from north-east

experi-generating a southward flow, while during the summer monsoon the wind blows from a

southern ditection creating a northward flow The circulation pattern in the Gulf of

Tonkin depends on the seasonal monsoon (winter and summer monsoon, respectively)

During the dry season the southwards flow dominates and is characterised by an clockwise rotating cell During the wet season there are two circulation cells whieh di-verge near the coastline of the Red River delta (sce Figure 2.9)

anti-Figure 2.0 Residual flow in the Gulf of Tonkin, based on the Vietnam National Atlas, 1906

(Gonree: van Maren 2004)

Residual flow inthe Gulf of Tonkin during the dry season (February) and the wet season

(August) is shown in Figure 2.9, based on the National Atlas of Viethann (1996) The 50,

200, and 2000 motors depth contours aro presented for reference

‘Mean wind velocity during the winter monsoon season is higher than that in the summertonsoon season The significant wave height (#1) is up to 3 metres during 10% of thetime in the winter and up to 2 metres during the summer, In August and September thenorth of Vietnam is often struck by typhoons and this can give a signifieant wave height

up till 6 m (Sundstrém and Sédervall, 2004) There exist records of 30 vears (1949-1988)

‘wave observations at Hou Dau (nearshore) station and 25 years wind measured data atBach Long Vi island, which locates 200 kilometres offshore Nam Dink coasts, The esti-mated average offshore wave heights are in the range of 1.2-14 metres for summer andabout 2.0 metres for winter The significant wave height changes from 6.0-8.0 metres inthe offshore to 3.0-5.0 metres in the nearshore, Mean wave periods are in the range of 8-12 seconds These yearly average wave height and period are valid at depths of about 17‘metres near the Hai Han beach,

a7