Data Mining and Equi-Accident Zones for US Pipeline Accidents

Abstract

Data mining is performed on the last 21 years of United States pipeline accident data to illustrate the trends in different pipeline accident types and their consequences, namely financial losses, fatalities, and volume of oil spill. An objective ranking is introduced to identify the states with most pipeline accidents, losses, fatalities, and oil spills. The influence of meteorological season, temperature, and precipitation on pipeline accidents is investigated. The contiguous United States (excluding Alaska, Hawaii, and Puerto Rico) is partitioned into six accident zones with equal amounts of accidents (equi-accident zones) and the most frequent pipeline accident type in each zone is identified. Among all the pipeline accident types, material/equipment/weld failure is found to be the most frequent, expensive, and environmentally unfriendly pipeline accident type. Pipeline accidents due to natural force damage increase dramatically during winter season. Among the six equi-accident zones, Zone 2, located in the southern United States, is the smallest zone with the highest density of accidents.