A multi-attribute Systemic Risk Index for comparing and prioritizing chemical industrial areas

https://doi.org/10.1016/j.ress.2011.10.002Get rights and content

Abstract

Measures taken to decrease interdependent risks within chemical industrial areas should be based on quantitative data from a holistic (cluster-based) point of view. Therefore, this paper examines the typology of networks representing industrial areas to formulate recommendations to more effectively protect a chemical cluster against existing systemic risks. Chemical industrial areas are modeled as two distinct complex networks and are prioritized by computing two sub-indices with respect to existing systemic safety and security risks (using Domino Danger Units) and supply chain risks (using units from an ordinal expert scale). Subsequently, a Systemic Risk Index for the industrial area is determined employing the Borda algorithm, whereby the systemic risk index considers both a safety and security network risk index and a supply chain network risk index. The developed method allows decreasing systemic risks within chemical industrial areas from a holistic (inter-organizational and/or inter-cluster) perspective. An illustrative example is given.

Introduction

The concept of ‘systemic risk’ is well known in the financial world where it is connoted with risks, which are common to an entire financial market and not to any individual entity thereof. Systemic risks also exist within the chemical industry. Although the nature of systemic risks (w.r.t. causes, prevention, etc.) is very different in the financial and the chemical sector, the potential consequences are in both cases devastating, both from a social as well as an economic point of view.

In the (petro)chemical industry, economies of scope, environmental factors, social motives and legal requirements often force companies to ‘cluster’. Therefore, chemical plants are most often physically located in groups and are rarely located separately. These clusters of chemical plants consist of atmospheric, cryogenic and pressurized storage tanks, large numbers of production installation equipment, and numerous pipelines for the transportation of chemicals and petrochemicals.

Clearly, such chemical industrial areas are characterized by reciprocal danger between equipment and infrastructures being part of the areas. As such, within chemical clusters intangible interdependencies between equipment and infrastructures may exist from a safety and security point of view. Every chemical installation represents a hazard depending on the amount of substances present, the physical and toxic properties of the substances and the specific process conditions. Hence, such installations present – to a greater or lesser extent – a danger to their environment (and thus to the other installations in the neighborhood). Besides losses of lives, both short and long term disruptions from accidents in the chemical industry have led to significant economic losses and environmental damage [1]. One type of accident particularly interesting in this regard is an escalating accident or a so-called domino effect, whereby one accident at one installation triggers another accident either at the same installation (temporal domino effect) or at another installation in the vicinity (spatial domino effect), leading to a major devastating accident. The reader interested in domino effects and domino accident prevention is referred to [2], [3], [4].

It is obvious that also strong tangible supply chain interdependencies do exist between the installations (and companies) composing a chemical industrial area. Supply chain interdependence is not limited to a single industrial area. Natural disasters such as the 1999 Taiwanese earthquake, 2005 hurricane Katrina, 2010 Icelandic volcano eruptions, but also large company accidents (2001 fire in the Phillips semiconductor plant in New Mexico, 2005 Buncefield oil storage depot disaster in the UK, 2010 explosion and sinking of the BP-operated oil rig ‘Deepwater Horizon’ 50 miles off the US-Louisiana coast, 2011 Japanese earthquake-tsunami disaster, etc.) have illustrated the cascading effects of major disruptions along the supply chain. Different risk events in the supply chain are linked to each other in complex patterns with one risk leading to another, or influencing the outcome of other risks [5] and are therefore intrinsic to supply chain management.

Although most companies tend to develop plans to protect against high frequency, low impact risks in their supply chains and tend to ignore high impact, low likelihood risks [6], disaster and disruption management have received increased attention during the last decade, both from a safety and security and from a supply chain point of view, respectively. Examples of this increased attention can be found in [7], [8], [9].

This paper builds upon recent research on domino accident prevention to construct a multi-attribute index for managing safety and security and supply chain related systemic risks. Section 2 provides an overview of current literature. Current safety indices used in safety management in the chemical and process industries are discussed together with state-of-the-art research on supply chain risk management. Compatible network representations are built in Section 3, whereas safety and security-, and supply chain indices for measuring systemic risks are constructed in Section 4. In the Section afterwards, both indices are forged into one user-friendly so-called Systemic Risk Index for comparing and managing systemic risks in chemical industrial areas. An illustrative example is given in Section 6. Section 7 briefly discusses the usefulness of our approach. The conclusions of this article are formulated in Section 8.

Section snippets

Safety and security management literature

Many safety indices have been developed for a number of different purposes in chemical industrial settings. They are extensively used for ranking various chemical installations based on the hazards these installations represent, possibly leading to accident scenarios such as fire, explosions, BLEVE, toxic releases, etc. Well-known examples are the Dow fire and explosion index F&EI [10], [11], Dow chemical exposure index CEI [12] and the Mond fire, explosion and toxicity index [13], [14]. Other

Safety and Security Network (S&S Network)

Reniers and co-researchers elaborated a methodology to represent a chemical industrial area as a weighted directed network for managing knock-on accident prevention. In this section we briefly explain how this approach can be adjusted to set up a safety and security network. For a more elaborate discussion on the methodological foundations of similar networks and applications using empirical data, the reader is referred to [30], [31].

In a so-called weighed graph, a variable represents the

Safety and Security Network Index (S&S Index)

As already mentioned in Section 3.1, the escalation threat that every vertex (installation) poses to every other vertex in the safety and security network can be represented by the Domino Danger Unit associated with the directed edge between every couple of vertices. Therefore, in case of the safety and security network, the network's resilience can be examined by investigating the network's typology using the domino danger units. We thereby assume that each potential path through the network

Multi-attribute model for comparing and ranking chemical clusters with respect to systemic risks

Investigating the safety and security network's and the supply chain network's typologies allows us to develop a systemic risk picture for both types of networks.

By using both the S&S Index and the SC Index, we can compare the behavior of different network configurations, and e.g. increase the reliability of supply chain key input flows by introducing redundancies, or decrease domino effects risks and threats by improving safety and security countermeasures.

It should be noted that these

Illustrative example

In this section, we show the usefulness of the developed method by applying it to a (hypothetical) example. To this end, we have artificially generated the data for four hypothetical chemical clusters. The data have been generated in the following way. First, we generate the installations randomly on the Euclidean plane between (0,0) and (1000,1000). All distances are calculated and normalized between 0 and 1 (dividing by 1000×sqrt(2)). Since the distance between two installations determines to

Conclusions

To investigate the systemic risk features of a cluster network, an industrial area is considered as a graph and is represented by its safety and security matrix and by its supply chain matrix. Subsequently, the Safety and Security Index and the Supply Chain Index of the network are calculated and the Borda algorithm is used to determine the relative position of chemical industrial areas with respect to their systemic risk behavior. This way, a mathematically sound methodology is developed to

References (42)

  • M. Maroño et al.

    The ‘PROCESO’ index: a new methodology for the evaluation of operational safety in the chemical industry

    Reliability Engineering and System Safety

    (2006)
  • G.K. Al-Sharrah et al.

    A new safety risk index for use in petrochemical planning

    Process Safety and Environmental Protection

    (2007)
  • C.T. Leong et al.

    Inherent safety index module (ISIM) to assess inherent safety level during preliminary design stage

    Process Safety and Environmental Protection

    (2008)
  • A. Tugnoli et al.

    Safety assessment in plant layout design using indexing approach: implementing inherent safety perspective. Part 1—Guideword applicability and method description

    Journal of Hazardous Materials

    (2008)
  • A. Tugnoli et al.

    Safety assessment in plant layout design using indexing approach: implementing inherent safety perspective. Part 2—Domino hazard index and case-study

    Journal of Hazardous Materials

    (2008)
  • G.L.L. Reniers et al.

    Knock-on accident prevention in a chemical cluster

    Expert Systems with Applications

    (2008)
  • G.L.L. Reniers et al.

    Managing domino effect-related security of industrial areas

    Journal of Loss Prevention in the Process Industries

    (2008)
  • G.L.L. Reniers et al.

    Continuously and simultaneously optimizing an organization's safety and security culture and climate: the Improvement Diamond for Excellence Achievement and Leadership in Safety & Security (IDEAL S&S) model

    Journal of Cleaner Production

    (2011)
  • S. Koch et al.

    Software project effort estimation with voting rules

    Decision Support Systems

    (2009)
  • W.E. Martin et al.

    An application of social choice theory to U.S.D.A. Forest Service decision making

    Journal of Policy Modeling

    (1996)
  • M. Zarghami

    Soft computing of the Borda count by fuzzy linguistic quantifiers

    Applied Soft Computing

    (2011)
  • Cited by (27)

    • Sustainability and risk – a review of energy security

      2021, Sustainable Production and Consumption
    • Ranking chemical industrial clusters with respect to safety and security using analytic network process

      2019, Process Safety and Environmental Protection
      Citation Excerpt :

      Supply chain interdependences also exist between installations, increasing the degree of connection, for better or worse, between companies in the cluster area. Natural disasters, but also large industrial accidents have illustrated the cascading effects of major disruptions along the supply chain (Reniers et al., 2012). Thus, safety and security do not stop at companies’ fences, on the contrary, an accident, deliberately caused or not, can easily trigger even more severe events in nearby companies.

    View all citing articles on Scopus
    View full text