Currently, the problems of environmental safety are facing society. The industry develops every year. In this critical economic situation, the oil industry is a stimulant for the economic sector in Azerbaijan. The level of development of this industry is also relevant due to other reasons: sociological, technological and features of the economy of Azerbaijan. The production activity of oil processing, concentrating harmful substances and energy, is a source of man-made danger and pollution of the natural environment. The risk management process mainly consists of three stages - risk safety analysis, risk assessment, which is carried out in comparison of calculated and actual risk levels, the so-called acceptable risk levels and the adoption of appropriate regulations and management decisions. One of the factors that should be taken into account when assessing risk and safety is to determine the necessary costs. Since these costs are paid directly to the company, they try to minimize them as much as possible, which reduce the accuracy of risk assessment. One of the objectives of the study is to determine the optimal value of the necessary costs. It is established that the less reliable the method, the lower is the cost of its implementation. The methodological basis of the work was scientific works on these problems of scientists-economists, mathematicians on safety and risk assessment at industrial enterprises. When developing the presented methodology, computational algorithms developed by Dow Chemical were used. This company has collected a large volume of material on accident statistics, taking into account damages. Based on the obtained and experimentally verified data, a system of indices has been developed, an assessment of various indicators for qualitative and quantitative risk assessment of oil refineries. The analysis of the technogenic danger of oil refineries makes it possible to determine ecological and economic losses and choose rational possibilities of acceptable risk. The required costs, depending on the level of risk, are determined based on an increase in the accuracy of calculating the probability of occurrence of the cause of risks. The article examines the relationship between the expected level of risk and economic losses during oil refining in separate technological units, which allow determining the required level of risk and the expected economic damage.

It is known that an increase in the volume of production in any sphere of production as a consequence increases the intensity of anthropogenic impact on the environment, including the atmosphere. This problem is especially typical for oil refineries. On the one hand, this is due to the complexity of oil refining processes and the release of various hydrocarbon compounds into the atmosphere during processing. Therefore, when implementing sustainable ecological and economic development in the Azerbaijani regions that are associated with oil refining, an increase in the volume of products should not create an increase in environmental impact. This problem is considered one of the factors directly affecting the economic situation of the enterprise, as well as creating certain environmental, social and economic problems for society as a whole. The purpose is to study the possibility of using the correlation-regression method in solving the problem of determining the presence of the decapling effect. The relationship between the volume of petroleum products produced and the amount of pollutants released into the atmosphere is random and is characterized by stochastic and statistical dependence. At the same time, on the basis of a mutual comparison of the methods of correlation and regression analysis and the method of production function, the effectiveness of correlation and regression analysis is determined. The methodological basis of the work was the scientific works of scientists-economists, mathematicians providing the definition of the decoupling effect at industrial enterprises. In the presented methodology, the obtained estimates of the performed in real time vary depending on the values of the set parameter. When developing the presented methodology, the STATISTIKA software package was used. This program was developed by StatSoft. Based on the obtained and experimentally verified data, the equation of linear regression dependence is determined. During the research, it was found that in some cases it is not possible to evaluate the statistical characteristics of a random variable or it is accompanied by serious errors. Therefore, when processing data, instead of mathematical expectations and variances, we used selective mathematical expectations and variances. The analysis of the ecological and economic situation of oil refineries based on correlation and regression analysis allows us to determine the presence of the decapling effect. The obtained correlation coefficient between the volume of production oil products and the volume of emissions into the atmosphere shows that there is no statistically significant relationship between these parameters. Thus, there is an achievement of the decoupling effect between the production of petroleum products and the volume of emissions of pollutants into the atmosphere.