PowerPoint-esitys

Document Sample
PowerPoint-esitys Powered By Docstoc
					                Presentation of the EU study (1997)
            “THE USE OF DECISION-AID METHODS IN THE
           ASSESSMENT OF RISK REDUCTION MEASURES
                 IN THE CONTROL OF CHEMICALS”

                      Dr. Joonas Hokkanen
             Consulting Engineers Paavo Ristola Ltd
                             Finland




Nov 2004            Joonas Hokkanen                   1
DEVELOPMENT OF A
RISK REDUCTION
STRATEGY. AN
EXAMPLE OF THE
PROCEDURE UNDER
THE EU ‘EXISTING
SUBSTANCES
REGULATION’.




   Nov 2004        Joonas Hokkanen   2
  The assessment of the risk reduction measures leads
           to the typical multicriteria problem




Nov 2004               Joonas Hokkanen                  3
       THE FIVE TYPICAL PROBLEMS IN THE DECISION MAKING PROCESS


                                                                                                                            g
                                                           d                                                       d    b



                 a               f                                                  a           f

                         e       g                                                          e   g
                 b                                                      g        b

                                                                    a
                 c       d       h                                              c           d   h
                                           c   f               b                                          c   f    e    a
                                                       e


                A. Choice problematic; help                                     B. Sorting problematic; help
                choose a "best" action                                          sort actions to the feasible
                                                                                ones and not feasible ones.




                                                                                                          c

                                                           c

                                                                                a           f
                                                                            b                         a       b d
                                     a d
                     a                                                            e g
                             f                                                  b                    e
                         e   g
                     b                                     e                    c       d   h

                     c   d   h                                                                        f           g h
                                                   f           gh



                  C. Ranking problematic without the                             D. Ranking problematic with the
                  intensity between the alternatives                             intensity between the alternatives.




                     E. Description problematic; Describing the variety of
                     valuations
                     different (weight combinations) that support the choice of
                     alternative
                     that


Nov 2004                                                            Joonas Hokkanen                                             4
                 Traditional MCDA-approach


    Criteria
    measures
       x
                              Decision     Best solution
                               model
                               u(x,w)           ?

      DMs’
    valuations
        w


Nov 2004                 Joonas Hokkanen                   5
     CRITICAL INFORMATION NEEDED FOR THE APPLICATION OF DIFFERENT
  PROCEDURES AND METHODS FOR A TRANSPARENT AND CONSISTENT DECISION
Class                 Subclass                            Critical information
Disaggregative        Procedures with no pre-ference or   Normalisation technique
procedures            criteria level information          Ordinal ranking
                      Procedures with some criteria       Standard levels
                      level information                   Comparison order
                                                          Criteria and alternative removal
                                                          Monetary valuation
                                                          Trade-offs
                      Procedures based on ordinal         Ranking order
                      preference inform-ation             Criteria and alternative removal
Aggregative methods   Methods based on utility theory     Value functions
                                                          Utility values
                                                          Weights
                                                          Pairwise comparison matrices
                      Monetary transformation methods     Transformation procedure

                      Scoring techniques                  Normalisation technique
                                                          Weights
                      Outranking methods                  Weights
                                                          Preference thresholds
                                                          Indifference thresholds
                                                          Veto thresholds
                                                          Ranking coefficients
                      Continuous mathematical             Reference points
                      programming                         Constraints used


Nov 2004                                   Joonas Hokkanen                                   6
CONNECTION OF THE PLANNING PROCESS AND ALTERNATIVE
             SUBSTITUTION ASSESSMENT
Fundamental steps of the             Alternative Assessment            Decision Aid tools
Planning Process
Definition of objectives             Definition of objectives          Expert judgment, Group
                                                                       Participation, Interviews etc

Generation of alternatives           Defining the alternative plans    Expert judgment, Group
                                                                       Participation, Interviews etc
Formulation of criteria /            Defining the impact of the        Expert judgement, Group
measures of effectiveness            alternative plans                 Participation, Interviews etc


Evaluation of alternatives           Evaluation of alternatives        Expert judgment, Group
                                                                       Participation, Interviews etc

Comparison of the alternatives       Comparing the alternative plans   MCDA Disaggregative
/Conflict solving                                                      procedures, partially and totally
                                                                       aggregative methods etc
                                     Planning how to reduce the
                                     potential nuisances and how to
                                     monitor the effects.
Selection of preferred alternative                                     MCDA Disaggregative
                                                                       procedures, partially and totally
                                                                       aggregative methods etc




   Nov 2004                                     Joonas Hokkanen                                            7
                  SMAA - approach


     Criteria
     measures
        x
                           Decision
                            model      Prospective
                            u(x,w)      solution



     Favorable
     valuations
           ?
Nov 2004             Joonas Hokkanen                 8
                   SMAA Weight Space Analysis

                                                                  a 1 =33%
•   Identifies the favorable
    weights Wi that make                              W1
    alternative i the preferred
    one                                                                         a 2 =38%




                                  Criterion 2
•   Acceptability index ai =                            x1             W2
    the expected volume of                              (3,9)
    Wi                                                                 x2
•   Central weight wic =                                             (7,7)
    centroid of Wi
                                                                                W3         a 3 =29%

                                                                             x3 (9,2)

                                                                                   Criterion 1




    Nov 2004                                    Joonas Hokkanen                                  9
                     Rank r acceptability indices

                                                               b 11 =33%
•   Rank acceptability indices
    are similarly computed for                                     b 21 =21%
    each rank r= 1,…,m
                                                                                 b 12 =38%
•   The resulting acceptability



                                  Criterion 2
                                                       x1                            b 22 =62%
    profile can be plotted and
    used for identifying                               (3,9)
    compromise alternatives                                          x2
                                                                     (7,7)
                                                                                          b 23 =18%

                                                                                                   b 13 =29%
                                                                               x3 (9,2)

                                                                                     Criterion 1




    Nov 2004                                    Joonas Hokkanen                                         10
Comparison between the aggregative multicriteria methods.
Feature                 Ordinal                          Cost-benefit                   Utility based methods             Outranking methods              Methods     based      on
                                                                                                                                                          mathematical optimisation
                        ranking of the                   analysis
                        criteria

Preference modelling    Ranking order of each            Monetary transformation        Utility function.                 Threshold model.                e.g. aspiration levels
                        alternative for each criterion

Uncertainty modelling   Uses threshold model.            Uses             probability   Uses                probability   Uses threshold model.           Uses                   probability
                                                         distributions.                 distributions                                                     distributions.

Compensation            Fully compensative.              Fully compensative.            Fully     or          partially   Fully     or        partially   Totally compensative.
                                                                                        compensative.                     compensative.

Transformation of the   Not used                         The decision makers must       The decision makers must          Not used.                       The decision maker must
basic data                                               accept a common type for       accept a common type for                                          accept a normalisation of
                                                         the monetarisation.            the utility function.                                             basic data.

Scales and scaling      No scaling necessary             The decision makers must       The decision makers must          No scaling is necessary, the    No scaling is necessary, the
                                                         accept the monetary scaling.   accept the utility scaling.       scales are to be considered     scales are to be considered
                                                                                                                          when     determining     the    when      determining    the
                                                                                                                          threshold values                aspiration levels.

Weights                 Represent     the     relative   Represent the willingness to   Represent     the     relative    Represent votes given to the    Represents       the    reference
                        importance of the criteria,      pay the cost.                  importance of the criteria,       importance of a certain         points
                        can be used to determine                                        can be used to determine          criterion, cannot be used to
                        trade-off coefficients.                                         trade-off coefficients.           compute       the   trade-off
                                                                                                                          coefficients.

Ranking                 To choose the ”best” one         Complete ranking of the        Complete ranking of the           Partial ranking i.e. the        To choose the ”best” one
                                                         alternatives                   alternatives                      incomparability is accepted.




          Nov 2004                                                          Joonas Hokkanen                                                                                  11
                                                                                  INFORMATION USED WHEN PREPARING THE DECISION
Table 3: Information used with different procedures and methods                   No Preference information                                                                                                        Preference information is used
and the type of the problem which can be solved.                                  used                                                                                                                                                                                                    Criteria values                                                                                                      Solving the
                                                                                                                                                                                                                                                                                          have to be                                                                                                           problems (see
                                                                                                                                                                                                                                                                                          transformed to                                                                                                       figure 3)
                                                                                                                                                                                                                                                                                          equal units




                                                                                                                                                                                                                                                                                                                                                               No transformation needed with criteria values

                                                                                                                                                                                                                                                                                                                                                                                                                Choose or sort the "best" alternative/s
                                                                                                                                                                                 Ranking order on some criterion




                                                                                                                                                                                                                                                       Cardinal weights of the criteria
                                                                                                                                                                                                                     Ordinal ranking of the criteria
                                                                                    Descroptive evaluation table

                                                                                                                   Reduced Evaluation Table




                                                                                                                                                                                                                                                                                                            Monetary transformation

                                                                                                                                                                                                                                                                                                                                      Utility transformation




                                                                                                                                                                                                                                                                                                                                                                                                                                                          Rank the alternatives

                                                                                                                                                                                                                                                                                                                                                                                                                                                                                  Describtion problem
                                                                                                                                              Dominance rules




                                                                                                                                                                                                                                                                                            Normalization
                                                                                                                                                                Standard level
Procedures with no preference or criteria level information
           Dominance analysis                                                      x                               x                          x                                                                                                                                                                                                                                                                     x
           Positional analysis                                                     x                               x                          x                                  x                                                                                                                                                                                                                                  x
Procedures with criteria level information
           Standard level analysis                                                 x                               x                          x                 x                                                                                                                                                                                                                                                   x
           Risk/Benefit analysis                                                   x                               x                          x                 x                                                                                                                                                                                                                                                   x
Procedures based on ordinal preference information
           Lexicographic analysis                                                  x                               x                          x                                  x                                  x                                                                                                                                                                                               x
           Elimination by aspects                                                  x                               x                          x                                  x                                  x                                                                                                                                                                                               x
Aggregative methods based on ordinal ranking of the criteria
           Agreed criteria                                                         x                               x                          x                                                                     x                                                                                                                                                                                               x
           Individual                                                              x                               x                          x                                                                     x                                                                                                                                                                                               x
Aggregative methods based on cardinal prefernce information
           Simple additive weighting method                                        x                               x                          x                                                                     x                                  x                                    x                                                                                                                       x                                       x
           Cost/Benefit analysis                                                   x                               x                          x                                                                     x                                  x                                                    x                                                                                                       x                                       x
           Simple Multiattribute Rating                                            x                               x                          x                                                                     x                                  x                                                                              x                                                                             x                                       x
           Analytic Hierrachy Process                                              x                               x                          x                                                                     x                                  x                                                                              x                                                                             x                                       x
           Outranking methods                                                      x                               x                          x                                                                     x                                  x                                                                                                       x                                                    x                                       x
Mathematical optimization
           e.g. methods based on aspiration level                                  x                               x                          x                                                                     x                                  x                                                                                                       x                                                    x                                       x
Aggregative method for solving the weights making each alternative the best one
           Stochastic MulttiaAttribute Acceptability Analysis, SMAA                x                               x                          x                                                                                                                                                                                       x                        x                                                                                                                    x




   Nov 2004                                                                 Joonas Hokkanen                                                                                                                                                                                                                                                                                                                                                                                                             12
THE LOGICAL FLOW FOR
APPLYING DIFFERENT
PROCEDURES AND
METHODS




 Nov 2004              Joonas Hokkanen   13
  MCDA (Multi Criteria Decision Aid) has many
  advantages:

  1.       It is open and explicit
  2.       The choice of objectives and criteria that any decision making
           group may make are open to analysis and to change if they felt
           to be appropriate
  3.       Scores and weights, when used, are also explicit and are
           developed according to established techniques
  4.       Performance measurement can be sub-contracted to experts,
           so need not necessarily be left in the hands of the decision
           making body itself,
  5.       It can provide an important means of communication, within the
           decision making body and sometimes, later between that body
           and wider community, and
  6.       Scores and weights are used, it provides an audit trail.




Nov 2004                           Joonas Hokkanen                          14
           www.ristola.com




Nov 2004       Joonas Hokkanen   15

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:3
posted:10/5/2012
language:Unknown
pages:15