Research On Analysis Method Of Complex Leakage Accident Based On Case-based Reasoning

Well leakage is one of the common risks in the drilling process. If not monitored and warned timely, it may lead to serious leakage risk. When the well leakage is handled inappropriately or untimely, it may also result in further deterioration of the situation and even cause other drilling accidents, which is a serious threat to the safety of drilling. However, well leakage cannot be completely eradicated with current technology, so it is necessary to research the pre-warning and processing methods of well leakage. At present the comprehensive mud logging is mainly used in China. This method can pre-warn the well leakage risk through engineer's monitoring whether the early warning index data has reached a certain threshold and its own experience. But warning accuracy and timeliness depends heavily on the subjectivity of the threshold and the engineer's professional quality as well as experience.Drilling specialists are dedicated to the use of artificial intelligence technology in real-time warning and processing of the well leakage in recent years, but most of the intelligent warning and processing methods are problematic in terms of the high cost in knowledge acquisition, the complex processing rules and self-evolution, so the warning and processing effect is not ideal. Case-Based Reasoning has great advantage in solving the problems with numerous sources, rich experience, incomplete information of internal mechanism which is not fully understood. The DrillEdge, a drilling risk identification system developed by Verdande, has been successfully applied in warning and processing some complex drilling accidents. In such context, this text, based on the early-warning and processing index, put forward the further research in complex well leakage accidents based on case-based reasoning, which is scientifically meaningful both to case-based reasoning itself and the analytic method of the complex accident.The method consists of the risk pre-warning and the optimal processing scheme of well leakage. It mainly includes three key technologies:Case Representation, Case Retrieval and Case Study. The whole working process is as follows:Firstly, classify the real-time index data of the pre-warning and the processing as a new case. Secondly, use the case retrieval technology to find the similarity between this new case and those in well leakage case base. Next, match the similar cases from the case base. Finally, pre-warn the risk in real-time and screen the optimal processing scheme with the use of similarity and relevant information from the similar case.In order to further verify the feasibility and scientificity of this method, this paper develops intelligent and real-time analysis software based on case reasoning with the support of C# language and SOL Server 2008 database. The simulation test of software effect is carried out using the well leakage cases in Sichuan basin. The test results show that this method can assist or replace the engineer to pre-warn the risk of well leakage and put forward the optimal processing scheme in real time. Compared with conventional methods, it is not affected by geology, drilling operations or human subjectivity. Therefore, this method has a wider application, stronger intelligence and better pre-warning and processing effect.