Modelling the accidental oil spills at potentially hazardous facilities in the Arctic zone of Krasnoyarsk Krai

Abstract

Intensified human activity in exploiting the natural resources leads to increasing technogenic load on Arctic’s fragile ecosystem. The boost in industrial facilities is associated with a growing number of stationary fuel reservoirs poorly monitored due to their considerable remoteness and extreme weather conditions. The 2020 emergency in the Arctic zone of Krasnoyarsk Krai exposed the lack of adequate methods for risk assessment and behaviour in case of accidents at potentially hazardous facilities. The existing methodologies for assessing the area of spill following an accidental depressurisation present significant limitations. Most methodologies are based on analytical models not taking into account the physics of processes. This work uses modelling with neural networks of oil spill at the potentially hazardous object located in the Arctic territory of the Krasnoyarsk Krai. The software used was neural network simulator NeuroPro, developed in the Institute of Computational Modelling of Krasnoyarsk Scientific Centre of SB RAS. For training the neural network there were used daily operational data on fourteen main vectors affecting the propagation rate. The neural network modelling of the accidental oil spill during the depressurization of one of the fuel tanks at potentially hazardous facilities in the Arctic zone in 2020 correlated perfectly with the real data.