Building vulnerability and human casualty estimation for a pyroclastic flow: a model and its application to Vesuvius

Pyroclastic flows clearly present a serious threat to life for the inhabitants of settlements on the slopes of volcanoes with a history of explosive eruptions; but it is increasingly realised that buildings can provide a measure of protection to occupants trapped by such flows. One important example is Vesuvius, whose eruption history includes many events which were lethal for the inhabitants of the neighbouring Vesuvian villages. Recent computational fluid dynamics computer modelling for Vesuvius [Todesco et al., Bull. Volcanol. 64 (2002) 155–177] has enabled a realistic picture of an explosive eruption to be modelled, tracing the time-dependent development of the physical parameters of a simulated flow at a large three-dimensional mesh of points, based on assumed conditions of temperature, mass-flow rate and particle size distribution at the vent. The output includes mapping of temperature, mixture density and mixture velocity over the whole adjacent terrain. But to date this information has not been used to assess the impacts of such flows on buildings and their occupants. In the project reported in this paper, estimates of the near-ground flow parameters were used to assess the impact of a particular simulated pyroclastic flow (modelled roughly on the 1631 eruption) on the buildings and population in four of the Vesuvian villages considered most at risk. The study had five components. First, a survey of buildings and the urban environment was conducted to identify the incidence of characteristics and elements likely to affect humanvulnerability, and to classify the building stock. The survey emphasised particularly the number, location and type of openings characteristic of the major classes of the local building stock. In the second part of the study, this survey formed the basis for estimates of the probable impact of the pyroclastic flow on the envelope and internal air conditions of typical buildings. In the third part, a number of distinct ways in which human casualties would occur were identified, and estimates were made of the relationship between casualty rates and environmental conditions for each casualty type. In the fourth part of the study, the assumed casualty rates were used to estimate the proportions of occupants who would be killed or seriously injured for the assumed pyroclasticflow scenario in the Vesuvian villages studied, and their distribution by distance from the vent. It was estimated that in a daytime eruption, 25 min after the start of the eruption, there would be 480 deaths and a further 190 serious injuries, for every 1000 remaining in the area. In a night-time scenario, there would be 360 deaths with a further 230 serious injuries per 1000 after the same time interval. Finally, a set of risk factors for casualties was identified, and factors were discussed and ranked for their mitigation impact in the eruption scenario. The most effective mitigation action would of course be total evacuation before the start of the eruption. But if this were not achieved, barred window openings or sealed openings to slow the ingress of hot gases, together with a reduction of the fire load, could be effective means of reducing casualty levels.