Frequency of damage by external hazards based on geographical information

External explosions can significantly contribute to risk of damage for industrial plants. External explosions may origin from other plants in the neighborhood, which store and operate with explosive substances, or from transport of such substances on road, rail, or water. In all cases, some accident is a necessary condition for a hazard. Another probabilistic element is the probability of ignition. If transport causes the explosion, the location of the accident will influence the consequences. If deflagration is involved, ignition will not necessarily occur at the place of the accident, but a cloud of a combustible gas-air mixture may develop, which will ignite at some distance depending on wind velocity. In order to avoid unnecessarily pessimistic approaches, geographical information can be used in addition to local weather statistics. Geographical information systems provide map material for sites, roads, rail and rivers on a computer. This information can be used to find frequencies of damage based on numerical integration or on Monte Carlo simulation. A probabilistic model has been developed. It is based on: - A joint probability density function for wind direction and wind speed, which has been estimated from local weather statistics, - Frequency of hazards for neighboring plants and various types of traffic, - Statistics on the amounts and types of explosive materials, - The model has been implemented using one numerical integrations method and two variants of Monte Carlo method. Data has been collected and applied for a nuclear power plant in Northern Germany as an example. The method, however, can be used for any type of plant subject to external explosion hazards. In its present form, it makes use of design criteria specific for nuclear power plants, but these could be replaced by different criteria. (orig.)