Dataset of NaTech events triggered by lightning within the Process Industry
Description
This study delves into the impactful realm of lightning strikes, a meteorological phenomenon that poses substantial risks to the process industry by potentially triggering technological disruptions. Despite constituting merely 13% of NaTech (Natural-Technological) events, lightning events hold significant consequences within the process industry. To comprehend this, a meticulous analysis of past lightning-triggered incidents was conducted, utilizing open-source industrial-accident databases to assemble a dataset comprising 689 records. The research not only builds upon prior studies concerning NaTech events triggered by lightning but also updates the information, encompassing events until the close of 2022. A systematic classification system was introduced, in adherence to NaTech risk field recent studies, enhancing the accuracy of information. Concurrently, adjustments were made in compliance with international regulations governing lightning strikes. The study involved an event tree analysis, incorporating recent conceptual updates regarding the propagation path of NaTech events. Bayesian structures were employed, revealing interdependencies and vulnerabilities among functional variables. Quantitative analysis disclosed that over 80% of these events involved Incidents or Loss of Containment, transpiring mostly during regular plant operations. Alarmingly, records indicated a lack of information regarding safety barriers, emphasizing the need for heightened focus on this crucial aspect. Seasonal patterns emerged, with 83.3% of incidents transpiring during spring and summer, primarily in the Chemical and Petrochemical sector. Storage equipment followed by electrical equipment and electrical devices were the most vulnerable components, constituting 40% and 21% of events respectively. Fire scenarios dominated at 56%, involving substances classified as physical hazards in approximately 70% of cases. Event tree analysis revealed the damage source in 65% of records, with potential categorization as major accidents, according to Seveso Directive criteria. Implementing Bayesian networks provided a robust framework, elucidating conditional probabilities governing critical dependencies associated with damage, including source, state, and final scenario. Insights from past incidents bolster industrial vulnerability awareness against lightning strikes. These insights are part of an article to be submitted to the Journal of Process Safety and Environmental Protection, entitled "NaTech triggered by lightning: Learning from past events within the process industry".
Files
Steps to reproduce
The methodological cornerstone was the data collection about industrial accidents caused directly or indirectly by a lightning strike. The data was mainly retrieved from open-source databases which re-ported data about past industrial accidents until the end of 2022. In addition, open-literature information was used in some cases to contrast and deepen the information available in the reports. The methodology was integrated under the principle of convergence, from previous research reporting industrial accidents triggered by diverse natural events. For the compilation of the raw data, several currently available open-source databases on industrial accidents were consulted, both from Europe (ARIA, eMARS, TAD IChemE, eNATECH) and the United States (NRC, CSB). Since the research scope is focused on identifying vulnerabilities against lightning strikes able to trigger a NaTech event within the process industry as critical infrastructure, then, the following necessary inclusion criteria were adopted to consider the records: The technological event should involve the release (or the potential release) of hazardous, triggered by lightning impact, defined as a single electrical discharge of atmospheric origin between cloud and earth consisting of one or more strokes. The event should be classified as an accident, incident, or loss of containment event, or near misses. To ensure relevant criteria from the chemical and process industry, the registered technological events occurred in the process industry or eventually in an industrial sector of interest that handles a significant amount of hazardous substances. They should be clustered within eight industrial macro-sectors i) chemical and petrochemical, ii) storage and warehousing, iii) power production, iv) bioprocess, v) water treatment, vi) transportation, vii) pipeline, and viii) manufacturing. Moreover, the interrogations to each database were carried out using the keywords “lightning”, “thunderstorm” or “flash” as queries, according to previous research. In addition, the keyword was logically combined with other keywords such as “establishment”, “facility”, “plant”, “industry”, “chemical” and “hazardous”. In some cases, the keywords should be translated into other languages according to the original database languages. Furthermore, scrupulous attention was paid to data extraction to avoid duplicate records existing in more than one database, in those cases the most precise and detailed incidents were chosen, complementing the information. The database structure was designed through labels and categories from information mainly categorical. Many labels are free entrance such as code, date, continent, country, and location/city, to which corresponded with specific data of the associated record. In contrast, some labels require a not evident nominal categorization, for which diverse criteria established in the literature were used. Further details about the references can be found in the article.