1. Introduction
Natural events are widely recognized as triggering factors of industrial accidents. Accidents initiated by the impact of a natural event on an industrial site are referred to as Natech (Natural event triggering a Technological disaster) scenarios [
1]. The concern related to these peculiar events increased in recent decades due to the occurrence of severe accidents, such as those following the Kocaeli earthquake in 1999 [
2], the hurricanes Rita and Katrina in 2005 [
3], and the Great East Japan earthquake and tsunami in 2011 [
4]. The number of severe natural events and disasters is growing over the years [
5,
6], making industrial sites more exposed to such events, thus increasing the frequency of Natech accidents [
7,
8]. Thus, growing attention is dedicated in the literature to Natech accidents, and several recent studies are contributing to building a clearer and complete understanding of the features of Natech scenarios. However, to date, the attention has mostly focused on intense natural disasters having a limited duration, such as earthquakes, floods, tsunamis, and hurricanes. For such events, the lessons learned were derived from the detailed analysis of relevant Natech accidents [
9,
10,
11], and vulnerability models were developed to evaluate the failure probabilities of equipment items, as in the case of earthquakes [
12,
13], floods [
14,
15,
16], and hurricanes [
17]. Moreover, for these categories of natural events, several methodologies for the Natech quantitative risk assessment are available in the literature, based either on the extension of conventional QRA [
18,
19,
20] or other innovative approaches [
21,
22]. Thus, the consolidated knowledge and quantitative approaches to risk assessment are present in the literature for Natech accidents triggered by such intense natural events.
Nevertheless, Natech accidents can be triggered in principle by any kind of natural event, even those characterized by “low” intensity, such as extreme temperatures, which caused a relevant number of accidents in the past [
7,
8,
23]. However, limited attention was devoted to date to the interactions with industrial installations of these types of events [
24,
25].
When considering extreme temperatures, two different scenarios may be recognized. The first is related to the impact on industrial installations of cold waves, causing low temperatures, snow, and the formation of ice. The opposite is related to heat waves, resulting in high temperatures, drought, and related events. As suggested by a preliminary study carried out by Ricci et al. [
26], these extremes may affect differently industrial installation, and a detailed analysis of the cascading events and resulting accident scenarios that may be triggered should be carried out separately. Recently, an in-depth analysis was carried out to assess the possible impacts of cold waves and industrial installations, considering both equipment items and safety barriers [
27]. Nevertheless, limited attention was devoted in the literature to the study of heat wave-related accidents. Thus, the present contribution aims at filling this gap by focusing on cascading sequences and Natech scenarios triggered by heat waves.
According to the Intergovernmental Panel on Climate Change [
28], climate change and global warming have increased the mean and the maximum ambient temperature, and similar trends are forecasted for future years. As a result, warm seasons are longer, and more severe drought conditions are experienced. These factors promote the conditions for development and the rapid spread of wildfires [
29,
30], which are one the most severe outcomes of this critical scenario. Indeed, extreme wildfires took place in recent years, such as those occurred in California in 2020 and 2021 [
31,
32,
33], Canada in 2015 and 2016 [
34], Portugal in 2017 [
35], and Greece in 2018 [
36], demonstrating the devastating consequences that the wildland–urban interface may suffer when hit by wildfires. In this framework, some studies focused on the hazards posed to industrial installations by wildfires spreading in wildland–urban [
37,
38] and wildland–industrial interfaces [
39,
40,
41]. Nevertheless, limited attention was devoted in the literature to the direct effects of heat waves and of extremely high ambient temperatures on industrial installation [
24,
25]. Actually, climate change is enhancing the occurrences and the intensities of such natural events. Moreover, heat waves usually impact extended areas, thus potentially impacting on a high number of installations.
In the present study, a dataset of past Natech accidents triggered by heat waves was collected and analyzed, aiming at improving the understanding of such accident scenarios and at drawing lessons learned that may contribute to preventing their recurrence.
Section 2 reports the methodology used for the data retrieval and describes the features of the dataset. Relevant aspects of the cause–consequences chain of cascading events triggered by heat waves are assessed in
Section 3. Lessons learned are discussed in
Section 4 identifying the specific pathways through which heat waves may affect industrial installation and providing recommendations that in perspective may be useful to prevent or mitigate Natech accidents caused by such events in the future. Some conclusions are drawn in
Section 5, remarking the threat posed by heat waves to industrial installations and confirming the need to properly manage the hazard deriving from heat waves in industrial sites.
3. Results and Discussion
3.1. Original Sources and Geographical Location of the Events
A total of 204 Natech accidents triggered by heat waves were collected based on the criteria described in
Section 2.
Table 1 reports the share among the original sources of the records included in the present study. The ARIA database represents the main source of information, providing around 70% of the events in the dataset (141 records). The second source is the NRC database, providing a total of 55 records included in the study (27% of the total). Although a limited number of records were collected from the other sources, a high level of detail, allowing the characterization of the cause-consequence chain of the accidents, is provided by the accident files obtained from the eMARS, the MHIDAS, and the TAD IChemE databases.
It is interesting to notice that the share among the original sources is quite different from that of the complete database of Ricci et al. [
7]. Indeed, in the generic Natech database, the main data source is the NRC database (over 85%), while the ARIA database provides only about 12% of the information [
7]. The different share among the original data sources is correlated to the differences in the geographical distribution of past accidents. Indeed, most of the accidents in the dataset occurred in Europe (144 records, 70.6% of the total) and in North America (59, 28.9%). The site of the accident was unknown in a single case. When considering the country, most of the accident records that occurred in North America took place in the United States of America, while France is the country where most of the past accidents recorded in Europe took place. Actually, these results are not surprising since most of the records in the ARIA database are based on events that occurred in France [
7,
48], while the NRC database only collects events that take place in the United States of America [
23,
49].
It is also relevant to compare the distribution of Natech accidents to that of natural events (i.e., extreme environmental phenomena that impact societies and the human environment [
53]). Data on the number of natural events recorded in specific geographical areas are available from several sources. In the present study, data from the EM-DAT database developed by the Centre for Research on the Epidemiology of Disasters [
5] were considered to compare the total number of natural events to that of Natech accidents in the last 30 years (1992–2022).
Figure 2 shows the ratio between Natech accidents and recorded natural events. In the figure, the data for generic Natech accidents in the database of Ricci et al. [
7] are compared to the specific data concerning Natech accidents caused by heat waves. The same categories of natural events that triggered the Natech accidents were considered when deriving the information on the number of natural events considered (i.e., all the natural events recorded and heat waves only). As shown in the figure, a different trend is obtained when considering North America and Europe. Data concerning North America suggest that heat waves are a minor cause of Natech accidents and the value of Natech accidents per natural event is far lower than 1. This result differs from the figures obtained for generic Natech accidents, that score an average of 4.7 accidents per natural event [
7], which is a credible value considering the high industrialization, the intensity, and the extension of natural phenomena such as hurricanes that affect some regions of North America. On the contrary, the ratio for Natech events caused by heat waves is higher than 1 in Europe, and it is higher than that calculated for generic Natech events, which is around 0.5. These results suggest a higher vulnerability of the European framework to accidents triggered by heat waves, also in the light of the specific features of the original sources. Indeed, the NRC database, which represents the main source of data for North America, tends to include accidents with low severity that are usually not recorded by the ARIA database, the source from which most of the European data are taken. The discrepancies between the two geographical areas may thus be even more significant when considering the exposure of industrial installation to heat waves. Actually, the ratio between Natech accidents and natural events calculated for Europe in the present study may somehow represent an underestimation of the actual value, given that low severity accidents may be overlooked and not reported in the ARIA database.
3.2. Technological Scenarios
Figure 3 shows the technological scenarios that occurred in the Natech accidents collected in the present study (see
Table A1 for the definition of the technological scenarios considered). For the sake of comparison, the figure also reports the data obtained for generic Natech accidents by Ricci et al. [
7] (i.e., all the Natech accidents recorded regardless of the natural events that triggered them).
When Natech events triggered by heat waves are considered, a fire took place in more than half of the events (102 records). A release with no further consequences occurred in a more limited number of cases (40, 19.6%), as well as environmental contamination (31 records, 15.2% of the total). Multiple scenarios occurred only in six past accidents (2.9%). These were fires and explosions (three records), explosions and toxic gas dispersions (two), and in a single case the combination of all these three scenarios.
These figures present significant differences with respect to those obtained in previous studies addressing generic Natech accidents [
7]. Actually, a much higher incidence of fire scenarios is obtained for Natech events triggered by heat waves: the share of fire scenarios is about five times higher with respect to generic Natech events. Conversely, a much lower incidence of releases with no further consequences and of environmental contamination is recorded for Natech events triggered by heat waves. This difference may be explained by considering the specific features of heat waves. Indeed, the extremely high temperatures experienced during heat waves may facilitate the ignition of the hazardous materials during a loss of containment and/or result in conditions where self-ignition of materials is possible.
3.3. Industrial Sectors Affected
The industrial sectors involved in past Natech accidents triggered by heat waves are shown in
Figure 4 (the definition of the industrial sectors considered in the present study is reported in
Appendix A).
Bioprocesses and Chemical and Petrochemical industries seem to be the more vulnerable sectors to such accidents, being involved in around 65 % of the recorded events. Other industrial sectors of concern are those related to manufacturing (26 records, 12.7%), storage and warehousing (15, 7.4%), and transportation via rail and road (14, 6.9%). A limited number of accidents occurred in other industrial sectors, such as power production, transport via pipeline, and water treatment.
Within bioprocesses, two main types of industries are considered according to the definition provided in
Table A2: the manufacture of food products and the solid waste treatment sector. The first is involved in about 15% of the events affecting bioprocesses. However, most of the accidents occurred in solid waste treatment plants and in landfills. Actually, these facilities are more involved in Natech accidents triggered by heat waves, suggesting the need for specific measures to reduce the vulnerability and mitigate this specific type of Natech scenario. The relevance of fire scenarios shown in
Figure 3 may explain the high number of accidents reported for bioprocesses and for the chemical and petrochemical sectors. Indeed, these industrial sectors are vulnerable to fire scenarios due to the presence of high quantities of flammable and/or combustible substances. In this framework, it is worth considering that the increase of the ambient temperature may lead to self-ignition phenomena, as well as to changes in solid composition, with the possible release of flammable volatile substances (e.g., see [
54]). Moreover, high ambient temperatures may exceed the flash point of combustible liquids, increasing the probability of ignition causing fire scenarios. Similarly, poor control of temperature profiles within warehouses may generate conditions favoring the thermal degradation and self-ignition of chemicals.
Another criticality concerning the waste treatment sector is related to the continuous variability in the composition of wastes, which makes the safety management for this sector even more complex. As an example, heat waves may lead to a reduction in the water content, thus increasing the probability of self-degradation and self-ignition. However, the temperatures at which such phenomena may occur are frequently unknown, due to the possible variability in the composition of the waste.
3.4. Equipment Items
Figure 5 shows the equipment items that were involved in the technological accidents triggered by heat waves. The classification of the equipment categories is reported in
Table A3.
The results shown in
Figure 5 allow a more effective characterization of the cause-consequence chain that occurred in the Natech events analyzed. Not surprisingly, storage equipment is the item category mainly affected in Natech accidents triggered by heat waves (61 records, 29.9% of the total). Actually, these equipment items are the more vulnerable ones considering their structural features.
Waste disposal represents the main critical section of the plant for the bioprocesses sectors. However, piles of waste materials and landfills were involved in Natech accidents triggered by heat waves. Clearly, piles of materials are not equipment items. Nevertheless, they were considered as solid waste storage systems in the present study, since they were involved in several past accidents and as a result were vulnerable to the hazards caused by high ambient temperatures.
Other categories of equipment items were affected by heat waves in a lower number of cases. Among these, warehouses are of specific concern, in particular when considering bioprocesses and in the chemical and petrochemical industry.
3.5. Direct Causes
Five main events were identified as direct causes of the past accident scenarios triggered by heat waves collected in the present study.
Figure 6 shows the number of events in which the direct causes identified triggered the technological scenario responsible for the final consequence of the Natech accident.
In over 40% of the Natech accidents collected in the present study, loss of containment was caused by an uncontrolled increase of the internal pressure, exceeding design limits and leading to equipment failure. Usually, the internal pressure increase is generated by the high ambient temperature leading to an increase in the vapor pressure and/or of the vaporization rate of the liquid phase within the equipment involved.
The second more important direct cause of technological scenarios identified in this category of accidents is the self-ignition of materials and substances (23 records, 11.3% of the total), directly causing a fire scenario. Chemicals and/or materials degradation due to high temperatures were also a result of concern, causing a relevant number of recorded accidents (14, 6.9%). In such scenarios, continuous exposure to the heat by solar radiation was usually the starter of exothermic decomposition, polymerization, and other undesired reactions, which led to the self-ignition of the substance/material. In other cases, the lens effect was identified as the cause of the accident (11, 5.4%). The lens effect occurs when a glassy material converges sunrays into a narrow area, heating-up and finally igniting the surrounding materials. Power outages, resulting in anomalous process conditions, were recorded as the direct cause of eight accidents (3.9%).
Figure 7 shows the distribution of the direct causes identified with respect to the equipment item involved in the accident. This information was available only for a total of 125 records. The figure evidences a non-homogeneous distribution of the direct causes of the accident for the different equipment categories considered. In particular, self-ignition and materials degradation are relevant causes of accidents in storage systems (tanks, warehouses, and piles of materials). More specifically, self-ignition has occurred in accidents involving only storage equipment (12.8%), waste disposal (23.8%), and warehouses (44.4%). Accidents caused by the lens effect occurred only in waste disposal and warehouses, with a probability much higher concerning the first one (47%) than the second (11%). The occurrence of ignition due to lens effects in these industrial activities was more relevant than the potential of glass materials to ignite wild vegetation reported in previous studies [
55]. However, solid waste materials and biomass in the presence of high temperatures and dry weather reduce their humidity content. This may have enhanced their susceptibility to ignition due to the lens effect. These results highlight the importance of proper management of storage equipment, storage methods, and storage procedures during heat waves.
Differently, in the case of process equipment, pipework, and machinery, where the inventory and residence times are much lower, such causes are not relevant, and the main direct cause of failure is internal over-pressurization. The pressure increase is the most relevant direct cause considering storage equipment (75%), and process equipment (60%). Moreover, a pressure increase was the only identified direct cause for road and rail tankers, pipework, valves, and instrumentation.
Power outages resulted in a specific direct cause of accidents for machinery (60%) and process equipment (40%), not affecting the other equipment categories.
It should be remarked that the results shown in
Figure 7 are not affected by the industrial sector in which the equipment categories are operating: similar trends were obtained for the different industrial sectors considered, showing no significant deviations from the overall data reported in the figure.
3.6. Substances Involved in the Natech Events
To better characterize the cause-consequence chain of Natech accidents triggered by heat waves, it is paramount to analyze the substances which were involved in the Natech accidents occurred. Actually, such substances are responsible for the technological scenario (i.e., fire, explosion, etc.) in the Natech accident. The technological scenario is triggered by the release of such substances (e.g., due to the damage of equipment items or storage vessels), by their ignition or by their self-degradation.
Information on the substances involved in the technological scenarios was reported only for about 65% of the events. Around 70 different substances were involved in the Natech events triggered by heat waves. Most of the substances were involved in a single or very few events. Only six substances and/or materials were involved in more than five events, as shown in
Table 2.
Wastes (not better specified) together with liquid hydrocarbons (i.e., fuel oil, crude oil, and other liquid hydrocarbons) are the substance categories more frequently involved in the collected set of past accidents (about 30% of events). This is in line with the two sectors more frequently involved in the set of accidents collected: bioprocesses (including waste treatment as shown in
Table A2) and the chemical and petrochemical industry.
Other two substances frequently involved in the past Natech events collected are ammonia (14, 6.5%) and propane (3.7%). Actually, these substances are frequently stored as liquefied gases under pressure at ambient temperature. Such storage conditions are inherently vulnerable to high temperatures, which lead to the increase of the vapor pressure of the liquid phase and thus of the internal pressure of the storage system, possibly causing the loss of containment of the substance. This is coherent also with the results obtained concerning the direct cause of accidents since internal pressure increase was found to be the most relevant cause of the recorded accidents.
Figure 8 shows the technological scenarios involving the four main substance categories present in
Table 2: Wastes, Liquid Hydrocarbons (including Crude Oil, Fuel Oil, and Other liquid hydrocarbons), Gaseous Hydrocarbons (including propane), and Toxic Gases (including Ammonia).
As shown in the figure, fire is almost the only scenario involving waste materials (94% of events). Toxic gas dispersion and environmental contamination took place only in a limited number of cases (3% of events each).
As for liquid hydrocarbons, scenarios that do not require ignition occurred in around 90% of the cases. These are environmental contamination (68%) and release with no other consequences (23%). Ignition leading to fire took place in around 6% of the events, and an explosion occurred in over 3% of the records in which a liquid hydrocarbon was released. Thus, the ignition probability of liquid hydrocarbons obtained from the analysis of the results reported in
Figure 8 (around 9%) is comparable to the figures obtained by Ricci et al. [
7] from the analysis of generic Natech accidents, as well as to the data reported for conventional accidents [
43].
In spite of the flammability of ammonia, almost all events involving this substance concern the dispersion of an unignited toxic cloud. Indeed, in a single record ammonia caused multiple scenarios which consisted of the explosion of the storage vessel and the spread of the toxic gas over a residential area.
When gaseous hydrocarbons are considered, the scenario most frequently reported was a release with no other consequence. Only in a single record a multiple scenario took place, specifically an explosion followed by a fire.
3.7. Severity of Consequences
The severity of the past accidents collected was assessed not considering near misses. Thus, a total of 192 records were included in the analysis. The severity was assessed in terms of damages to humans (fatalities and injuries) and economic losses. Notably, such information was available only for a limited number of records due to the lack of detail of the original sources (nine records reporting damages to humans and six records reporting economic losses). Nevertheless, a total of 243 injuries and 4 fatalities occurred in Natech accidents triggered by heat waves, and in total the economic losses amount to over EUR 25 million.
Figure 9a shows the results obtained for damage to humans, and
Figure 9b shows those related to economic losses. As in the study of Ricci et al. [
7], four classes of severity were defined and used to classify the data. In case an accident resulted in multiple effects (e.g., fatalities and injuries), the more severe consequence category was applied.
For the sake of comparison,
Figure 9 also includes the results obtained for generic Natech events by Ricci et al. [
7].
Figure 9 shows that Natech accidents triggered by heat waves seem to have a slightly higher severity with respect to the overall figures obtained for generic Natech events, even if in absolute terms the probability of damages to humans and economic losses is still rather low. This result is somehow surprising, since heat waves are considered a natural event of minor intensity with respect to hurricanes and earthquakes and are typically overlooked in the quantitative risk assessment of Natech accidents [
24,
25]. This may be possibly caused by a scarce awareness concerning this category of events, leading to the absence of specific actions aimed at their prevention and/or mitigation.