Spatial risk assessment in case of multiple nuclear release scenarios

https://doi.org/10.1016/j.seps.2019.06.006Get rights and content

Highlights

  • •

    Assessing the impacts of a nuclear accident requires more information than the contamination level registered in a given geographical area.

  • •

    Our approach allows to establish a risk rating for different geographical units for different assets, under different attributes and accident scenarios.

  • •

    The information provided can be used for multilateral negotiations and collective decision-making processes: it can be considered as a common basis from which start designing policies.

  • •

    We nevertheless adopt a client’s perspective providing the IRSN (Institut de Radioprotection et de Sureté Nucléaire) with the necessary information that their “clients” typically require.

Abstract

This paper presents an approach aimed at assessing multiple criteria spatial risk, where several methods are used for decision aiding purposes, with an application on marine nuclear releases. The case study simulates a post-accident analysis evaluating the impact upon the bay of Toulon of an accidental nuclear release. The problem is characterized by the presence of spatial features, multiple criteria describing the involved assets and uncertainties represented through multiple release scenarios and their corresponding probabilities.

Introduction

This work is part of a larger project aimed at developing theoretical and practical tools aiding to synthesize multiple criteria spatial risks in case of multiple nuclear release scenarios. A literature review with relevant papers on the integration of multiple criteria decision analysis tools in spatial decision problems until 2006 can be viewed at http://publish.uwo.ca/∼jmalczew/list.htm. Despite the existing literature aiming to understand the processes governing the fate of radionuclides in the environment, [[1], [2], [3]], we note that the concentration of a given isotope is a necessary but not sufficient information for making informed decisions. Let us consider the example of two geographic zones: the first one is characterised by an average concentration level and very important economic and environmental assets while the second one is highly contaminated but does not present any economic or environmental relevance. Clearly, the involved stakeholders will be differently sensitive to the impacts in both geographic plots.

Our case study deals with simulated releases from a nuclear submarine at the bay of Toulon, where one of the most important bases of the French Navy is located. In case of a nuclear accident, the incumbent Prefect (the regional authority) needs synthetic information to support decisions, such as banning certain economic activities, setting a new water management policy at each relevant zone or impeding the access to specific areas. The IRSN1 is in charge of a project aimed at improving models predicting dispersion and assessing the impact of radionuclides in the environment, see https://www.irsn.fr/FR/Larecherche/Organisation/Progra-mmes/Amorad/Pages/projet-Amorad.aspx#.XJeYzi17RQI. Several papers have been proposed to address the problem of nuclear risk assessment, see Refs. [5,22,25,27,29]. Most of previous works rest upon operational research analyses and very few papers addressed the post-accident risk assessment from a multiple criteria decision analysis point of view, see Refs. [4,19]. In order to provide supplementary post-accident management tools allowing to evaluate environment and economic impacts, we have developed an approach in which data associated to assets involved in the bay are paired with maps displaying the concentration level of a given isotope generating criteria maps. Each map describes the impact of a release concentration for a given criterion. We then use a multiple criteria aggregation procedure generating impact maps taking into account all assets. The final step consists of aggregating uncertain information over release scenarios (release positions, sea conditions, …) through an outranking approach. Our case study serves as a template that can be extended to other release events and geographical areas.

The originality of our work stands on the way we structured and modeled a practical issue, starting from the raw question “How can we evaluate the impact of a nuclear accident, similar to that of Fukushima, in the marine area?” The practical case was offered by the bay of Toulon, due to the presence of nuclear submarines in its port, characterised by the presence of multiple assets and two levels of spatial decomposition. In this paper, we propose the models used to assess the impact of a nuclear release on each asset involved, in case we are interested in identifying the most impacted assets or areas with respect to each asset, as well as to evaluate the global impact taking into account all considered assets.

The paper is organised as follows. Section 2 describes the case study including different decompositions of the area of interest and the associated data. In Section 3, we introduce the main theoretical concepts used in this work. We present in Section 4 the construction procedure of the criteria functions characterizing and evaluating the Bay. In Section 5, we show the results of the multiple criteria aggregation and the aggregation of release scenarios. We end up with a discussion. Several appendices provide additional details about this work and its results.

Section snippets

Case study

The area of interest is the Bay of Toulon (in what follows, we will use the Bay to refer to it), where a major basis of the French naval force is located, including nuclear submarines, besides being a densely inhabited area with important economic activities. Thus, there is a possibility of major negative impacts in case an accidental nuclear release takes place. Two features are identified in this study:

  • •

    Multiple impacts over different assets characterising the Bay.

  • •

    Uncertainties relative to

Multiple criteria decision analysis

In Section 2.1, we described the Bay of Toulon as a rich area where several assets are involved and can be impacted in case of a nuclear release. Our first objective is to define functions, which we shall call criteria, allowing us to assess the impact on each asset at each geographic unit. Each function evaluates a geographic unit from a single perspective. In consequence, we shall associate with each criterion a map evaluating the impact on the corresponding asset.4

Construction of criteria

The multiple criteria problem at hand adopts a rating formulation in which we consider the four criteria reflected in Table 6, with scales referring to the raw impact of a nuclear accident at each geographic unit. All criteria considered to evaluate the Bay are based on water quality through the concentration of cesium in water. Hence, the criteria will measure the impact of a given concentration on the assets involved at each geographic zone.

We start by presenting the typology of impact

Multiple impacts

We finally consider the case with multiple criteria. Recall that for each scenario (βi,RPj), which occurs with probability pij=qi×rj, we obtain four criterion maps:

  • •

    Fishing: gPe(βi,RPj)

  • •

    Fish farming: gFf(βi,RPj)

  • •

    Seagrass Posidonia: gSp(βi,RPj)

  • •

    Tourism: gT(βi,RPj)

The aim of this section is to aggregate effects due to:

  • •

    multiple criteria.

  • •

    uncertainty.

In the first part of this section, we shall solve the multiple criteria problem. In the second part, we aggregate uncertainties by considering scenarios as

Discussion

The paper presents an approach to assess the impact of a nuclear accident on the Bay of Toulon. This work can be extended to handle the case of different release accidents that are spatially and temporally very close by only updating the concentrations of radioactive substences in each geographic unit. The study still valid also if the accidents are not spatially close. However in case they are not temporally close, the results will be biased by the short terms impacts before the second release.

Conclusion

We have presented an approach to assess spatial risks, in cases characterised by the presence of several assets, spatial characteristics and uncertainties over the accident parameters (mainly the release position and sea currents). The developed approach is illustrated through an application of nuclear releases in the marine environment. The methodology aims to assess the impact of a nuclear accident at a geographic space (in our case the Bay of Toulon) as part of a post-accident analysis. In

Acknowledgment

The authors would like to thank the French National Agency Project AMORAD (ANR-11-RSNR-0002) for supporting this work and Cost-Action IS1304 for supporting a short term scientific mission at Instituto de Ciencias Mathematicas on expert judgment.

Oussama RABOUN is a Ph.D. candidate in computer science at LAMSADE, PSL University, Université Paris Dauphine, where he graduated as a master in operational research and decision aiding. He worked for three years in IRSN the French institute of expertise and research in radioprotection and safety of nuclear installations. During this three years, he intervened in a project “AMORAD” on developing decision aiding tools for environmental risk assessment in case of nuclear accidents in the marine

References (30)

  • D. Bouyssou et al.

    Evaluation and decision models with multiple criteria: stepping stones for the analyst

    (2006)
  • S. Bussotti et al.

    Morphological analysis of Posidonia oceanica flowers from a reef formation and a shallow water meadow of the Ligurian Sea (north-western Mediterranean)

    Vie Milieu

    (1998)
  • S. Chakhar et al.

    Gis-based multicriteria spatial modeling generic framework

    Int J Geogr Inf Sci

    (2008)
  • J.M. Devaud et al.

    UTADIS: une methode de construction de fonctions d’utilite additives rendant compte de jugements globaux

  • C. Dufresne et al.

    Wind-forced circulation model and water exchanges through the channel in the Bay of Toulon

    Ocean Dynam

    (2014)

    • Cited by (0)

    Oussama RABOUN is a Ph.D. candidate in computer science at LAMSADE, PSL University, Université Paris Dauphine, where he graduated as a master in operational research and decision aiding. He worked for three years in IRSN the French institute of expertise and research in radioprotection and safety of nuclear installations. During this three years, he intervened in a project “AMORAD” on developing decision aiding tools for environmental risk assessment in case of nuclear accidents in the marine area. His research interests include subjects such as multiple criteria decision analysis, preference modeling, artificial intelligence, and decision theory.

    Eric Chojnacki is researcher on risk modelling. He is an IRSN expert in statistics. He co-leads the axis risk & territory of the federation of research on the environment ECCOREV and is a teacher in statistics at the Aix Marseille University. He participates in numerous projects of research GNRC (evaluation of the uncertainty around the number of cases of leukemia attributable to the nuclear installations), RDT PRIME (construction of multiple criteria indicators of post-accidental nuclear damage), ANR CLARA &AMORAD (methods helping the decisions making concerning the maritime pollution), UMS & BEMUSE & PREMIUM (OCDE/ CSNI programmes which take into account the uncertainties in the nuclear safety studies), and works on the development and the implementation of methods which would help making decisions.

    Céline DUFFA is an IRSN radio-ecologist with 15 years of experience. She completed her PhD in environmental geosciences at Marseille University in 1998. She worked on environmental monitoring programs and development of operational models for radionuclide transfers for several years. She is currently working in Marine studies, sampling and research programs on French coasts and in Fukushima marine area, decision support tools development, and marine dispersion and transfer to organisms modelling. She is in charge of the development and the implementation for the IRSN crisis Center of the operational marine dispersion model for French marine areas. She used to manage environmental expertise studies. She also used to manage PhD and MSc students. One of them is currently working on processes that occur in estuarine areas - desorption of radionuclides in order to implement river source terms to marine dispersion models. Another current research activity deals with study of trophic transfer of Cs, and an ongoing program deals with measurements of contaminants in marine water - phytoplankton and zooplankton (collaboration with French Universities). More generally, she used to participate to various international working groups (Europe - IAEA). She is leader of the Marine Radioecology working group of the Alliance (European Radioecology network). She is author and co-author of more than 20 journal articles.

    David Rios Insua (Madrid, 1964) is AXA –ICMAT Chair in Adversarial Risk Analysis at ICMATCSIC, member of the Spanish Royal Academy of Sciences and Professor of Statistics and OR at URJC (on leave). He currently holds the scientific direction of the H2020 project CYBECO, in cyberinsurance and cyber risk management. He is scientific advisor of Aisoy Robotics and has received numerous awards as scientist and entrepreneur, including recently the prestigious De Groot Prize from ISBA. He has published 14 books and more than 130 refereed articles in his areas of interest which include decision analysis, risk analysis and Bayesian statistics and ther applications.

    Alexis Tsoukias (Greece, 1959) is a CNRS research director at LAMSADE, PSL University, Université Paris Dauphine. He holds (1989) a Ph.D. in Computer Science and Systems Engineering from the Technical University of Torino (Italy) where he also graduated engineering studies. His research interests include subjects such as multiple criteria decision analysis, non-conventional preference modeling, policy analytics, applied non-classical logics, ordinal mathematical programming, artificial intelligence, and decision theory. He is the co-author of two books and more than 80 journal articles and book contributions. He has been vice president of ROADEF (the French OR society) as well as President of EURO (the European association of OR societies). In 2007 he was the coordinator of the European COST Action IC602, Algorithmic Decision Theory, funded within the FP7. He served the Research Administration in several positions, including the National Committee of the CNRS, the Scientific Council of INERIS and the Scientific Board of the National Institute of Computer Science of the CNRS. He was the director of the LAMSADE until 2018 and is the national coordinator of the Policy Analytics Research Network (GDR 3270, www.gdr3270.fr). Besides teaching several post-graduate classes in Paris and worldwide, he is the coordinator of the Peace studies Master at Université Paris Dauphine. He occasionally practices decision support, mainly in the area of public policy. He has been invited to more than 30 Universities worldwide. Personal web page: http://www.lamsade.dauphine.fr/∼tsoukias.

    View full text
    © 2019 Elsevier Ltd. All rights reserved.