Advancing Hazard Assessment of Energy Accidents in the Natural Gas Sector with Rough Set Theory and Decision Rules

Marco Cinelli,Miłosz Kadziński,Matteo Spada,Peter Burgherr,Grzegorz Miebs

doi:10.3390/en12214178

Abstract

The impacts of energy accidents are of primary interest for risk and resilience analysts, decision makers, and the general public. They can cause human health and environmental impacts, economic and societal losses, which justifies the interest in developing models to mitigate these adverse outcomes. We present a classification model for sorting energy accidents in the natural gas sector into hazard classes, according to their potential fatalities. The model is built on decision rules, which are knowledge blocks in the form of “if (condition), then (classification to hazard class x)”. They were extracted by the rough sets method using natural gas accident data from 1970–2016 of the Energy-related Severe Accident Database (ENSAD) of the Paul Scherrer Institut (PSI), the most authoritative information source for accidents in the energy sector. This was the first attempt to explore the relationships between the descriptors of energy accidents and the consequence (fatalities). The model was applied to a set of hypothetical accidents to show how the decision-making process could be supported when there is an interest in knowing which class (i.e., low, medium, high) of fatalities an energy accident could cause. The successful use of this approach in the natural gas sector proves that it can be also adapted for other energy chains, such as oil and coal.

Highlights

Societies worldwide rely on energy to satisfy many of their needs, among others, cooking, warming, cooling, transportation, and electricity generation [1]
The last consolidated and validated database version for natural gas was used, including accidents from 1970 to 2016, and was employed to develop the classification model based on the decision rules
The results are presented in two sections, distinguishing relevance of attributes (Section 3.1), and the classification model based on the decision rules, including the classification of new energy accidents to hazard classes (Section 3.2)

Summary

Introduction

Societies worldwide rely on energy to satisfy many of their needs, among others, cooking, warming, cooling, transportation, and electricity generation [1]. Maintaining a constant provision of energy is a technological and political challenge, since there are continuous disruptions that can affect its reliable and efficient supply. They can be initiated by different causes, such as man-made (e.g., lack of maintenance), technological (e.g., collapse of an infrastructure), and natural (e.g., earthquake, flood), resulting in events such as, for example, explosions, fires, and release of toxic substances. These disruptions cause health, environmental, economic, and social impacts [2], such as fatalities, injuries, evacuees, ban on consumption of food, release of toxic substances, and economic losses [3,4].

Methods

Results

Conclusion