Assessing Accident Risks of Energy Technologies and Energy Infrastructure Networks Disruption and Recovery Processes

Abstract

Today’s societies and economies depend heavily on reliable supply of energy. In the past decades, a number of catastrophic events triggered by natural hazards, technical failures or intentional attacks have influenced the entire energy-related business. Consequently, analyzing such events has become a high priority for several stakeholders engaged in the safe and secure provision of energy. Since zero risk does not exist, meaning that adverse events could always occur, resilience is a promising concept to plan and build reliable energy systems. A systematic, in-depth, literature review on resilience assessment of energy infrastructure was conducted for the purpose of this thesis in order to identify the main trends and gaps. The findings show that the overarching goals of most of the existing resilience frameworks are twofold, namely to contribute to the prevention and minimization of adverse consequences, and to support a quick recovery after a disruption. The Future Resilient Systems (FRS) program builds on a resilience framework comprising enabling, cognitive and biophysical functions. This thesis addresses the biophysical functions involving the disruption and the recovery. Disruptive events, such as accidents, have been investigated using historical accident information. This requires, however, the availability of comprehensive and up-to-date data. Therefore, the Paul Scherrer Institute (PSI) established in the late 1990s the ENergy-related Severe Accident Database (ENSAD) that since then has been continuously updated. A key challenge is how to efficiently and comprehensively keep the database up to date, considering the vast and continuously growing amount of information that is especially available from full-text news archives and other unstructured data sources. This doctoral thesis presents and applies a novel procedure to systematically collect information in a semi-automated manner from a multitude of primary sources. The results demonstrate that the developed procedure outperforms earlier attempts of collecting information, in terms of comprehensiveness and completeness. The accidents recorded in ENSAD have been used in the past to assess selected risk indicators describing accident frequencies and consequences. However, the relationship between accident risk and specific indicators measuring, for instance, infrastructure quality, and health, safety and environment (HSE) aspects, has not been investigated yet. In this thesis, a multiple regression approach was used to identify potential accident drivers. The findings of this analysis indicate that the countries’ development status has in most cases a significant positive impact on the accident and fatality rate, and that countries with higher infrastructure quality and political stability perform better. The system’s response to a disruptive event, involving the disruption and the recovery, can be investigated with resilience assessment. A widely accepted approach to assess the resilience of critical infrastructure is the establishment and simulation of a system performance measure throughout a disruptive event. To simulate the system’s response, a comprehensive model of the European natural gas network was assembled from open-source information. For the general flow modelling of the network, the maximum flow approach from the complex networks’ theory was employed, and additionally natural gas consumption was geographically allocated. This allowed the introduction of a flow-based system performance measure. On a European level, the seismic hazard and technical failures, using ENSAD accident data, are taken into account. Two distinct disruption measures are introduced, namely disruption impact mapping and supply grade mapping, which allowed identifying and quantitatively assess critical pipelines in the network. In the context of resilience assessment, in addition to the disruption model, a recovery of network components is needed. However, previous studies generally lack recovery assessment, in particular in the case of natural gas networks. Therefore, in this thesis, recovery strategies are developed, simulated and compared. The application of the developed simulation approach at the European scale is challenging because modelling of recovery processes requires detailed, location-specific information. Therefore, the simulation model of the natural gas network was downscaled to the wider area of Leipzig (Germany). The network response is simulated subsequent to a potential flood event, because flood events caused high damage on the Leipzig network in the past. The analysis performed in this thesis demonstrated that intuitive post-event strategies boosting the recovery time (i.e. "opportunistic") do not necessarily outperform other recovery strategies. Furthermore, a comparison of pre-event and post- event strategies was carried out to systematically evaluate disruption and recovery processes. The results show that it makes sense to focus on pre-event strategies that are identified for specific pipelines to prevent a high system performance loss and to boost the subsequent recovery of the infrastructure. Overall, this thesis advances the current state-of-the-art in risk and resilience assessment of energy infrastructure. On the one hand, it provides insights on drivers of energy-related accidents. On the other hand, it was studied how to assess resilience of a natural gas network at different geographical scales. Ultimately, this can support energy infrastructure operators and authorities to plan and build safer and more resilient energy infrastructure.