Post on 27-Mar-2015
Contribution of multi-criteria decision aid methods for natural risk analysis and
management studies
Myriam M. MERAD1,3, Thierry VERDEL2, Romuald SALMON3, Bernard ROY1
1 Laboratory for Analysing and Modelling Decision-Aid Systems (LAMSADE)- Université Paris-Dauphine. Place du Maréchal de Lattre de Tassigny. 75775 Paris Cedex 16 (France).
2 LAboratory of Environnement, GeOmechanic and Structures (LAEGO). Ecole des Mines de Nancy. Parc de Saurupt. 54042 Nancy Cedex (France).
3 National Institut of Environnement and Industrial Risk (INERIS). Parc Technologique Alata. 60555 Verneuil-en-Halatte (France).
Introduction
Risk = Hazard Vulnerability (sensitivity) of the element at risk
Fig.1. Rooms and pillars exploitation
Fig. 2. Underground Instabilities: Subsidence
Fig. 3. Consequences of mining subsidence on constructions
The concept of «Risk Study»Risk management consists on :
Contextual analysis. Risk analysis.
Risk control.
The “Risk Study” : is a « decisional problem » ; is a multi-criteria problem ; can be handled using “collective expertise”.
Each step represents a “project” that we decide to
call “Risk Study”
Role of the « analyst » :
To contribute to the collective expertise. To propose an approach for modeling « Risk studies ».
Analyst
Subsidence
due to
underground activities
Risk Study
Collective expertise
MCDAM Project management
Organisation science
Methods and tools
Stakeholders (Actors) 1
Decision objects 2
Criteria 4
Problematics 3
Aggregation Procedures5
« Risk Study » is a « project » carried out within a « organization » ...
Multi-Criteria Decision Aid Method (MCDAM)
« Risk Study » is a decisional problem ...
The level of the decisional problem 1
The level of available information and knowledge
2
The Importance of the internal and external constraints
3
The dominant culture 4
The criticity of the context 5
1 2 3 4 5
1
2
3
4
5
A help for risk
management
The aim of this approach
To be able to practice a « Contextual analysis »A four Groups « Risk Study » typology
Rea
sona
ble
do
ubt
N
o u
ncer
tain
tyT
ota
l unc
ert
ain
ty
Operational Tactical Strategical
Level of decisional problem
Knowledge and
information
Type B
Forecast
Type D
Negotiate
Type C
Explore
Type A
Communicate
A description of the “type C”
Identifying and understanding the risk are the finalities of this type of « Risk Study ». The « actors » involved are of the « analytical » type ; they tend towards determinism even if their preferences are not set a priori. In this type of study the actor ’s needs and expectations are a help to structure the decision making problem.
Observations :
- No a priori preferences.
- Often a single actor (routine and analytical operations).
- Mixed information.
- Mainly technical.
- Tendency to determinism.
Requirements :
- Structuring the decision making problem.
To be able to practice a « Risk hierarchization »Proposition of some PMCAs for each « Risk Study » typology
Rea
sona
ble
do
ubt
N
o u
ncer
tain
tyT
ota
l unc
ert
ain
ty
Operational Tactical Strategical
Level of decisional problem
Knowledge and
information
Type BType D
Type C
Type AELECTRE (I, II, III, IV, TRI)NAIADE
PROMETHEE I /IIFuzzy
conjunctive/Disjunctive methodMELCHIOR
ORESTE ; REGIMEMartel et Zaras
Methods like ELECTRE (I, II, III, IV, TRI)
NAIADE
Fuzzy maximinFuzzy weighted
sum TOPSISMAVTUTA
SMARTMAUT
AHP ; EVAMIX
Fuzzy conjunctive/Disjunctive method
Martel et Zaras Method
Some suggestions concerning Aggregation procedures (MCDAPs):
- Preference structures (S, R) seem of interest. For “Type C”, the “incomparability” relation R lets the different actors ask themselves about the pertinence of the the “Risk Study” problem structure.
- PMCAs that use both quantitative and qualitative information.
[G1]. Are the stakeholders numerous or not? [G2]. Which cognitive procedure (Comparison, etc.) do the decision makers use ? [G3]. What is the decision reference problematic (ranking, choice, etc.)? [G4]. What information (quantity and quality) do we have? [G5]. What level of compensation the decision maker wish to obtain? [G6]. What are the basic assumptions of the MCAD one has ? [G7]. Is there a software that can handle with the principles of the PMCA to choose?
General questions : Hwang et Yoon (1981), Teghem et al. (1989), Orzony (1992), Laaribi et al. (1996), Guitouni et Martel (1998)
Some recommendations for Risk studies of the type C :
To be able to identify a change in the decisional context in the case of a long-term Project
Three levels of concern in decision are identified :
The level of the decisional problem 1
The level of available information and knowledge
2
The Importance of the internal and external constraints
3
The dominant culture 4
The criticity of the context 5
Importance of technical dimension
Decision impact
Information
Global
Very importa
nt
Weak Precise
Local
National
TacticalStrategical
Operational
A Case study : Risk of subsidence induced by
underground works in the North-East of France (Lorraine region)
Beginning of the Study: 1997. Extention: 120 km 30 km. Zones under building or infrastructures. Imperfect knowledge. « Making of » of Mining Risk Prevention Plan (MRPP).
A description of the study
The expected results
Land use proposal.
A Hazard zoning according to (3) urbanization constraints.
Risk zoning in four risk classes.
Multi-Criteria Decision Aid Method (MCDAM)
INERIS : Research and expertise activities. GEODERIS : Expertise group. GISOS : Scientific group. Local Expert committee. Ministries Associations of defense. Mayors.
1. Stakeholders (Actors)
1. The level of the decisional problem
Importance oftechnical
dimension
Decisionimpact
Information
Tactical
Operational
The study in1997
Crisis group management
The study in 2003
Planning and analysis group
The direct and/or indirect actors implied in the risk analysis study have changed : the socio-political dimension is less important that in 1997.
Type 2
Type 3
Type 1
Type 2
Type 3
Type 1
Homogenious zones which consists of many pillars and many galeries (431 zones).
2. Decision objects :
3. Problematics :
To sort the zones to one of the 4 risk classes.
Class 2
Class 3
Class 1
Class 4
zi
To sort the zones to one of the 3 urbanization constraints classes.zi
Shape of the zone to hierarchized
Presence of fault
(bar).τ1-
H0,25k
0 : Large pillars.
10 : Small pillars or tentering.
20 : Small irregular pillars .
ScoreQuali .Sensitivity to waterC1.5ScoreQuali .Size, shape of pillarsC1.4
ScoreQuali .Pillars superpositionC1.3
ScoreQuali .Presence of faultC1.2barQuant.Stress on pillarC1.1
Direction ofrisk
increasing
UnitTypeCriteriaN°
hectareQuant.Zone ExtentC2.4
cm/mQuant.Deformation (Am /H)C2.3
cmQuant.MaximumsubsidenceC2.2
mQuant.DepthC2.1
ScoreQuali .Networksvulnerability
C2.5.5
ScoreQuali .Constructionworksvulnerability
C2.5.4
ScoreQuali .Railwaysvulnerability
C2.5.3
ScoreQuali .Roadsvulnerability
C2.5.2
ScoreQuali .Constructionvulnerability
C2.5.1
4. Criteria
Size and shape of pillars « C1.4 »
Stress on pillar « C1.1 »
Which MCDAPs to choose for the Lorraine case?
Problematics: sorting P and ranking P. Experts commitee. Finalities: Installing monitoring devices ;
urbanization constraints.
Some specificities of the study …R
easo
nabl
e do
ubt
N
o un
cert
aint
yT
otal
unc
erta
inty
Operational Tactical Strategical
Level ofdecisionalproblem
Knowledgeand
information
Type BType D
Type C
Type AELECTRE (I, II, III, IV, TRI)NAIADE
PROMETHEE I /IIFuzzy
conjunctive/Disjunctivemethod
MELCHIORORESTE ; REGIME
Martel et Zaras
Methods like ELECTRE (I, II, III, IV, TRI)
NAIADE
Fuzzy maximinFuzzy weighted
sumTOPSISMAVTUTA
SMARTMAUT
AHP ; EVAMIX
Fuzzy conjunctive/Disjunctive method
Martel et Zaras Method
ELECTRE (I, II, III, IV, TRI)
NAIADE
PROMETHEE I /II
Fuzzy conjunctive/Disjunctive
method
MELCHIOR
ORESTE
REGIME
Martel et Zaras
Choice of the ELECTRE method
Type C
Some other observations ...3. Constraints More hardware. Geographical Information Systems (GIS). A relaxation of time constraints.
4. Dominant culture
Geotechnics.
5. Criticity of the decisional context
Crisis (1997) to Post-crisis (2003).
This makes it possible to be conscious of the group dynamics:
consensual or directed conclusions ;
a rupture of the dynamics of the innovation.
Conclusions and remarks
What are the contributions of the MCDAM?
Reintroduces the « analyst and the experts » in the modeling process.
Considers the various aspects of risk: (transdisciplinarity).
Show that « risk studies » aims action. Proves that complex problems are « decidables ».
Specifies the limits between « expertise » and « decision ».
MCDAMRisk Study
Helps to formulate the decision problem: organization and formalization.
Manages the interface « Analyst/Expertise (Decision aid)/Decision maker »:
SatisfactionSatisfaction Needs: To formulate the decision
problem ; to communicate; etc.
Modeling and conclusions.