A comprehensive risk assessment method for Christmas tree combined with a multi source information fusion algorithm and an improved HAZOP method

Abstract

The Christmas tree faces significant safety risks during installation and requires extremely high reliability during operation. In this study, a comprehensive risk assessment method combined GA-BP neural network and improved HAZOP method is established, which is the first application used in the field of risk analysis for offshore oil equipment. The sample risk data during the installation and operation of the China’s first domestically produced Christmas tree is extracted through multi-source fusion, and the possibility and severity levels of risk occurrence are classified and judged. By training and testing the model it shows that the average error rate of the proposed GA-BP multi-source information fusion method is only 5.10%. Through risk data extraction and evaluation, the failure probability when the Christmas tree lowering into water is found to be the highest, with a critical importance coefficient of 0.24. No high-risk failure modes are found during the production process, but there are 36 moderate risk points. Deep water gas well field testing shows the theoretical judgment is consistent with the actual discovery, and with using of hydrate inhibitor injection the blockage risk at the nozzle of the Christmas tree is alleviated, which verifying the accuracy of the method. The method has improved the risk warning system for the safe operation of the Christmas tree, and can provide technical support for similar subsea oil and gas production equipment.