A model for estimating the influence of chlorine scattering at its emergency releases into the atmosphere at a potentially hazardous object in the city of Krasnoyarsk

Abstract

Introduction. The article considers the risk of occurrence and development of an emergency situation caused by the occurrence of a landscape fire and the transition of a fire to technological buildings with further depressurization of containers containing chlorine. One of the threats to the city of Krasnoyarsk is chemically hazardous facilities that have the task of providing life support to the population and are located in complete isolation. These objects do not have a road connection with the coastline, which makes it practically impossible to use forces and means designed to respond to operational events at these objects in a timely manner. Goals and objectives. The aim of the study was to simulate the conditions of chlorine scattering during its accidental releases into the atmosphere and to identify the dependencies of the scattering parameters on the technological features of the release, weather conditions, as well as the characteristics of the environment where the release occurs. Methods. To simulate an emergency situation at a water treatment plant, the method of simulation modeling using the TOXI+Risk software product was used, and the method of neural network forecasting using the Scikit-Learn library in the Python programming language was used. Results and discussion. The simulation results demonstrated the possibility of using neural network modeling to solve the problem of short-term forecasting of the areas of dispersion of a chemically dangerous substance (chlorine). The analytical method and the neural network method are compared. Proposals have been developed to reduce the potential risk of an emergency. Conclusions. The use of a neural network model makes it possible to increase the speed of calculating the concentrations of AHS at various points in space in comparison with the use of a traditional integral model, as well as to assess the potential danger of scattering AHS in the event of destruction of the tank in the presence of a terrain model. However, the considered neural network model can predict the concentration exclusively in the training ranges of weather conditions. The combination of neural network and integrated models makes it possible to solve the problems of industrial safety under any circumstances. Key words: emergency chemically hazardous substance, emergency, chlorine, risk, threat, simulation modeling, forecasting, neural network model, analytical model.