APPROXIMATION OF POPULATION DISTRIBUTION UNDER AN ACCIDENT RELEASE PLUME FOR RADTRAN ACCIDENT RISK ANALYSIS OF COMPLETE ROUTES

Previously, semi-automated population tabulating tools have been developed at SNL as additions to the capabilities of a commercial geographic information system (GIS). One of these tools tabulates population within 0.8 km (1/2 mile) of the route centerline, in increments as short as 1 km, for input to RADTRAN incident-free dose calculations. The other tool tabulates population under individual isopleths of a dispersion plume extending from a hypothetical accident in a user-specified wind direction; it has been useful for analysis of consequences and risks associated with limited numbers of specific accident sites and conditions. A practical means of determining the distribution of population perpendicular to extended route segments, for RADTRAN accident-risk analysis, was not available. A method, that employs the inherent capabilities of the GIS, for tabulating population within bands parallel to both sides of a chosen portion of a shipment route (e.g. across a single state) is presented. The widths of the bands were chosen to equal the lengths of the ellipses describing a set of 15 isopleths commonly employed in RADTRAN accident-risk analysis. Population data were tabulated in this manner for potential routes across several western and mid-western states. They were compared with the corresponding incident-free population data normally entered in the RADTRAN accident-risk model. To evaluate the effects on risk calculations, the two sets of data were used as inputs to RADTRAN for the route segments investigated. Results revealed that the distributions of population density perpendicular to the route centerline vary noticeably from state to state. RADTRAN calculations for each state, with both sets of data, did not reveal a consistent bias and generally agreed within a factor of two. In conclusion, a change to the distributed population model is not necessary for suitable accuracy in evaluating accident risks relating to entire shipment routes. This new model promises to be most useful to analysts who need to consider populations at distances from a route that are comparable to plume extent.