era lecture 8 case studies

Upload: tino-diaz-iii

Post on 07-Apr-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/6/2019 ERA Lecture 8 Case Studies

    1/35

    Environmental Risk Analysis: Methodsand Applications

    '2008/2009Lecturer- Prof. Eugene Levner

    .8LectureOperations Research Tools in ERA.

    Case studies.

  • 8/6/2019 ERA Lecture 8 Case Studies

    2/35

    OUTLINE

    1. Descriptions of Two Environmental Problems

    2.1

    The Jordan River Problem. 2.2. The Dead Sea Problem.

    2. Two OR models 3.1. Risk mitigation planning for the Jordan River Problem

    (the facility layout and multiple-choice knapsack problems)

    3.2. Environmental risk minimization for the Dead SeaProblem (the multi-portfolio choice model).

  • 8/6/2019 ERA Lecture 8 Case Studies

    3/35

    Environmental Risk Definitions

    Risk is a likelihood that a course of actions (a lack of thereof) will

    result in an undesired event (US EPA 1998,2002).

    Environmental Risk is defined as a two-dimensional array

    consisting of: (1) a probability of a threat to human health, to the

    natural environment - air, water, and land - upon which life

    depends, and to health of flora and fauna, and (2) a magnitude of

    losses (Levner and Proth 2003,2005, Ganoulis and Levner,2007).

  • 8/6/2019 ERA Lecture 8 Case Studies

    4/35

    Qualitative Risk MatrixQualitative Risk MatrixAmount of PollutionAmount of Pollution

    & Probability of Damage& Probability of Damage

    Probability

    of Damage

    20 40 100

    0.1

    0.3

    0.5

    0.7

    0.9

    Impact = Amount of

    Pollution

    The matrix serves to rank the risks: the green tier denotes low level, grey acceptable, yellow - high, red very high. The matrix has the capability toevaluate the effectiveness of risk mitigation measures; white ell ipsescorrespond to three different situations defined by three different risk-aversionstrategies: a passive strategy leaves the risk level very high, a moderate policy

    decreases it to high, while an active strategy makes it acceptable.THIS COLORINGIS OUR FIRSTMAIN ASSUMPTION

  • 8/6/2019 ERA Lecture 8 Case Studies

    5/35

    Integrated Eco-Risk Estimation

  • 8/6/2019 ERA Lecture 8 Case Studies

    6/35

    Integrated Eco-Risk Index

    R=7j=1,, 57r=1,, RwjrRjr,

    where j is index of ecological risk classes j=1,5(human

    health, crops, animals, nature, infrastructure), and r is index of

    risk subclasses (age, diseases, professions, areas, etc.)

    wjris weight, or importance

    RandRjrare damage value (in physical or monetaryunits, or rating scale)

    [THIS SUMMATION IS OUR SECOND MAIN ASSUMPTION]

  • 8/6/2019 ERA Lecture 8 Case Studies

    7/35

    .1

    1.5. Integrated Risk-Cost Analysis using a

    combination of the environmental

    supply chain and house-of-risks.

    Modeling with the help ofOR tools

    (Knapsack problem)

  • 8/6/2019 ERA Lecture 8 Case Studies

    8/35

    ouse o s souse o s s

    for defining integrated risk magnitudefor defining integrated risk magnitude

    Weights

    0.8

    0.05

    0.05

    0.05

    0.05

    Risk R1

    Risk R3

    Risk R4

    Risk R5

    Risk R2

    Absolute risk value

    Costs

    Transport Consumers Irrigation SewageWater

    Quality

    Water

    QuantityFood ExposureRisk factor

    1

    Risk factor

    2Risk factor

    M

    Rows depict Risk classes. Columns Risk factors, Risk sources

  • 8/6/2019 ERA Lecture 8 Case Studies

    9/35

    An Illustrative Example

    An ecological map of Tenerife, Canarian

    Islands, designed together with Dr.David Alcaide Lopes de Pablo,University of La Laguna, Spain

  • 8/6/2019 ERA Lecture 8 Case Studies

    10/35

    Results of computationsResults of computations

  • 8/6/2019 ERA Lecture 8 Case Studies

    11/35

    2. Descriptions of Two

    Environmental Problems in Israel 2.1 The Jordan River Problem.

    2.2. The Dead Sea Problem.

  • 8/6/2019 ERA Lecture 8 Case Studies

    12/35

    The Jordan River ProblemThe Jordan River Problem

    In modern times the waters are 70 to90% used for human purposes andthe flow is much reduced. Moreover,the river is heavily polluted and in itslower part, just raw sewage andrunoff water from agriculture are

    flowing into the river. Most pollutedis the 60-mile downstream stretch -a meandering stream from the Sea

    of Galilee to the Dead Sea.

  • 8/6/2019 ERA Lecture 8 Case Studies

    13/35

    The Jordan River ProblemThe Jordan River Problem

    In the early 1960s, the Jordan Rivermoved 1.3 billion cubic meters (46billion cu ft) of water every year fromthe Sea of Galilee to the Dead Sea.

    But dams, canals and pumpingstations built by Israel, Jordan and

    Syria to divert water for crops anddrinking have reduced the flow bymore than 90 percent to about 0.10

    billion cubic meters (3.5 billion cu ft).

  • 8/6/2019 ERA Lecture 8 Case Studies

    14/35

    The Jordan River ProblemThe Jordan River Problem

    Environmentalists say the practice hasalmost destroyed the river'secosystem.

    The Jordan River will disappear ifnothing is done soon. More than halfof it is raw sewage and runoff water

    from agriculture. What keeps theriver flowing today is sewage -Friends of the Earth, Midddle East.

  • 8/6/2019 ERA Lecture 8 Case Studies

    15/35

    Specific Objectives:

    To design the water balance for all main water sources andprovide a list of water saving strategies in the Jordan RiverBasin, (including innovative technologies for waste water

    treatment, alternative agricultural and irrigation techniques,desalination and water treatment stations, intensiverainwater harvesting, etc.)

    Using the supply chain and House-of-Risks approach,evaluate the social, economical and ecological risks ofdifferent water resources utilization scenarios, at present

    and in the future. Provide a comprehensive ORbased optimization model as a

    flexible tool for scientifically motivated and fair waterallocation between all the water stakeholders in the JordanRiver Basin.

    The Jordan River ProblemThe Jordan River Problem

  • 8/6/2019 ERA Lecture 8 Case Studies

    16/35

    The Dead Sea ProblemThe Dead Sea Problem

    Main Threats to the Dead Sea - water pumping from Lake Kinneret and

    the Yarmouk River for water supply hascreated a water deficit about 800 millioncubic meters per year;

    - industrial solar evaporation ponds atChemical Works are responsible for about

    20% of the total evaporation of Dead Seawaters;

    -additional threats come from theuncoordinated tourism industry, hotels,transport, road building, etc.

  • 8/6/2019 ERA Lecture 8 Case Studies

    17/35

    The Dead Sea ProblemThe Dead Sea Problem

    The overal goal is:

    To develop an OR-based multi-

    criteria optimization model forintegrated management of waterresources for the Dead Sea

    Basin

  • 8/6/2019 ERA Lecture 8 Case Studies

    18/35

    RiskRisk--Oriented Optimization ModelsOriented Optimization Models Which risk-mitigating strategies to select? Which water treatment facilities to use?

    Which water/wastewater technologies to use?in order

    TO MINIMIZE INTEGRATED REGIONAL RISK IMPACTSTO MINIMIZE INTEGRATED REGIONAL RISK IMPACTSTO MINIMIZE TOTAL COSTSTO MINIMIZE TOTAL COSTS

    TO MINIMIZE UNCERTAINTYTO MINIMIZE UNCERTAINTY ((variance of returnsvariance of returns fromfromthe portfolio of chosen strategies, facilities andthe portfolio of chosen strategies, facilities and

    technologiestechnologies xx ))..

    under budgetory, technological, resource, legal andunder budgetory, technological, resource, legal andsocial constraints.social constraints.

    A General Framework forA General Framework forTwo ProblemsTwo Problems

  • 8/6/2019 ERA Lecture 8 Case Studies

    19/35

    PROBLEM FORMULATIONPROBLEM FORMULATION(Facility Location Problem)(Facility Location Problem)

    Input A set ofwater stakeholders or demand

    points D,

    A set of water/wastewater treatmentfacilities Fwith facility creating cost fi ,

    Connection cost Cij(not necessarily obeythe triangle inequalities),

    Environmental risk imposed Rij.Output A subset of facilities Fd An assignment of demand points from D to

    facilities in F

    ijijxR

  • 8/6/2019 ERA Lecture 8 Case Studies

    20/35

    Facility Location ProblemFacility Location Problem

    Objectives

    (1) Minimize the total cost (facility

    building + connections)

    (2) Minimize the environmental risk involved

    )involvedrisksntal(environmemin

    costs)connectioncostscreating(min

  • 8/6/2019 ERA Lecture 8 Case Studies

    21/35

    Asimple case: The bi-criteria FLPThe ILP Formulation

    FiDjyx

    FiDjyx

    Djxts

    xRMin

    yfxCMin

    iij

    iij

    Fi

    ij

    Fi Dj

    ijij

    Fi Dj Fi

    iiijij

    e

    !

    ,}1,0{,

    ,

    1..

    Each demand point should be assigned to one facility.

    Demand points can only be assigned to created facilities.

  • 8/6/2019 ERA Lecture 8 Case Studies

    22/35

    PROBLEM FORMULATIONPROBLEM FORMULATION(Facility Location Problem)(Facility Location Problem)

    Input A set ofwater stakeholders or demand

    points D,

    A set of water/wastewater treatmentfacilities Fwith facility creating cost fi , Connection cost Cij(not necessarily obey

    the triangle inequalities),

    Environmental risk imposed Rij.Output A subset of facilities Fd An assignment of demand points from D to

    facilities in F

    ijijxR

  • 8/6/2019 ERA Lecture 8 Case Studies

    23/35

    Facility Location ProblemFacility Location Problem

    Objectives

    (1) Minimize the total cost (facility

    building + connections)

    (2) Minimize the environmental risk involved

    )involvedrisksntal(environmemin

    costs)connectioncostscreating(min

  • 8/6/2019 ERA Lecture 8 Case Studies

    24/35

    Asimple case: The bi-criteria FLPThe ILP Formulation

    FiDjyx

    FiDjyx

    Djxts

    xRMin

    yfxCMin

    iij

    iij

    Fi

    ij

    Fi Dj

    ijij

    Fi Dj Fi

    iiijij

    e

    !

    ,}1,0{,

    ,

    1..

    Each demand point should be assigned to one facility.

    Demand points can only be assigned to created facilities.

  • 8/6/2019 ERA Lecture 8 Case Studies

    25/35

    Asimple case: The bi-criteria FLPThe ILP Formulation

    FiDjyx

    FiDjyx

    Djxts

    xRMin

    yfxCMin

    iij

    iij

    Fi

    ij

    Fi Dj

    ijij

    Fi Dj Fi

    iiijij

    e

    !

    ,}1,0{,

    ,

    1..

    Each demand point should be assigned to one facility.

    Demand points can only be assigned to created facilities.

  • 8/6/2019 ERA Lecture 8 Case Studies

    26/35

    Indices

    k= water stakeholders in the SC in aregion, k= 1, K.

    i = risk classes, i= 1, , I.

    j = stressors (sources of waterpollution),j= 1,, J.

    m = index of different water saving andrisk mitigating strategies andtechn l ies m =1 M.

    AnotherAnother simplesimple case: The bicase: The bi--constrained knapsackconstrained knapsackThe ILP FormulationThe ILP Formulation

  • 8/6/2019 ERA Lecture 8 Case Studies

    27/35

    Variables

    xim - technological/management choice

    variable :x

    im =1, if the m-th riskmitigation strategy is chosen fordecreasing i-th risk;

    xim =0, otherwise, m=1,, M.

    The problem is to choose the set of thecounter-pollution strategies, facilitiesandactivities so as to minimize the

    integrated risk,or, equivalently,to

    AnotherAnother simplesimple case: The bicase: The bi--constrained multiconstrained multi--choice knapsackchoice knapsack

    The ILP FormulationThe ILP Formulation

  • 8/6/2019 ERA Lecture 8 Case Studies

    28/35

    MC2KP

    Maximize D = 7i=1,,N 7m=1,,MiDimxim(there are N risk groups, and Mi items in each group, totally M1 +M2 ++MN = M)

    Subject to: 7i=1,,N 7m=1,,Mipimxim p0,(1)

    The first constraint requires that the total probability of

    damage does not exceed the allowed levelp0 .

    7i=1,,N 7m=1,,Mibimxim B0, (2)The second constraint imposes the bound on budget

    B0.

    7m=1,,Mixim ki , i=1,,N (3)The third group provides the balanced choice between

    different groups (classes) of environmental risks.

  • 8/6/2019 ERA Lecture 8 Case Studies

    29/35

    The RiskThe Risk--based Water Resources Management Problem for thebased Water Resources Management Problem for the

    Dead Sea as aDead Sea as aMMultiulti--portfolio Choice Problemportfolio Choice Problem

    Given a m-dimensional vector budget(amount of money available toinvest, along with other tools, suchas human and information resources)and a list of management strategies1,, n requiring investment, how

    can the vector budget be optimallydivided among the various waterresources management strategies for

    the saving of the Dead Sea?

  • 8/6/2019 ERA Lecture 8 Case Studies

    30/35

    The RiskThe Risk--based Water Resources Management Problem for thebased Water Resources Management Problem for the

    Dead Sea as aDead Sea as aMMultiulti--portfolio Choice Problemportfolio Choice Problem

    Denote by xij the amount of thejthcomponent of the m-dimensionalvector budget allocated to

    management strategy i, for i=1,,n,j=1,, m. Then the nxm matrix x,that we call a multi-portfolio, is amulti-dimensional decision variable

    for the problem. A goal of theoptimisation process is tocharacterise and find the optimumportfolio of management strategiesfor the develo ment of the Dead

  • 8/6/2019 ERA Lecture 8 Case Studies

    31/35

    The RiskThe Risk--based Water Resources Managementbased Water Resources ManagementProblem for the Dead Sea as aProblem for the Dead Sea as aMMultiulti--PortfolioPortfolio

    Choice ProblemChoice Problem

    Let the total return from portfoliox be therandom variablev(x), and (x) = theexpected value of returnv(x) from portfolioxin a pre-specified period. It is a measure ofthe long term average return per period fromthe portfolio.

    Note that in this approach, the returnv(x) isa vector function whose components reflect

    separate economic, technical, environmentaland social returns (benefits, welfare) that arequantitatively estimated by using the utilityfunctions for each stakeholder.

  • 8/6/2019 ERA Lecture 8 Case Studies

    32/35

    based Water Resources Managementbased Water Resources Management--The RiskThe RiskProblem for the Dead Sea as aProblem for the Dead Sea as aPortfolioPortfolio--ultiultiMMChoice ProblemChoice Problem

    Finding an optimum portfolio of waterresources management strategies shouldmaximise the expected return andminimise the environmental risk; inother words these objectives should beachieved simultaneously.

    Finding an optimum portfolio ofintegrated water resources managementstrategies is therefore a multicriterion

    optimisation problem.

  • 8/6/2019 ERA Lecture 8 Case Studies

    33/35

    based Water Resources Managementbased Water Resources Management--The RiskThe RiskProblem for the Dead Sea as aProblem for the Dead Sea as aPortfolioPortfolio--ultiultiMMChoice ProblemChoice Problem

    Following a financial risk management approachproposed by Harry Markowitz in 1952, we mayassume that the environmental risk of a portfolio can

    be quantitatively characterised by thevariance ofreturns from the portfoliox. Our Markowitzean approach is applicable to water

    resources management and extends the basicMarkowitz model in that (1) the variable portfoliox isthenxm matrix rather than an-dimensional vector of

    variable assets, and (2) each objective function (i.e.,the return and risk) is, in fact, a vector of severalfunctions,for different risks and differentstakeholders.

  • 8/6/2019 ERA Lecture 8 Case Studies

    34/35

    based Water Resources Managementbased Water Resources Management--The RiskThe RiskProblem for the Dead Sea as aProblem for the Dead Sea as aPortfolioPortfolio--ultiultiMMChoice ProblemChoice Problem

    The present multi-portfolio methodology

    is more complicated and computationallyless tractable than the classicalMarkowitz model. However, it allowspowerful mathematical methods of

    financial risk analysis (see, e.g.Rockafellar and Uryasev 2000 ) to beexploited for measuring and minimising

    environmental risks .

  • 8/6/2019 ERA Lecture 8 Case Studies

    35/35

    BibliographyBibliography1. K.-H. Elster, E.G. Golshtein, E. Levner, et al., Modern

    Mathematical Methods of Optimization, Akademie Verlag,Berlin, 1993, 416 pp.

    2. E.Levner, I. Linkov and J.-M. Proth, Strategic Management

    of Marine Ecosystems, Springer, Berlin,2005, 313 pages,ISBN 1-4020-3157.

    3. E.Levner, J.Ganoulis, I.Linkov, Y. Benayahu, Multi-objective risk/cost analysis of artificial marine systemsusing decision trees, in I. Linkov (ed.), Risk ManagementTools for Environmental Security, CriticalInfrastructureand Sustainability, Springer, 2007.

    4. H. M. Markowitz,Portfolio selection,Journal of Finance, Vol. 7, No.1,

    pp.77-91,1952.

    5. Rockafellar, R. and Uryasev, S., Optimization of conditional Value-at-

    Risk,Journal of Risk, No. 2, pp. 2142,2000.