Intelligent kick warning based on drilling activity classification

Kick is one of the risks that frequently occur in the drilling process, so efficient and precision warning is very important. The missed warning will cause a blowout, which seriously affects the safety and economics of the well site. This study establishes a new kick warning model based on drilling activity classification. With the proposed model, a kick can be detected quickly and accurately. Firstly, the relationship between different drilling activities and drilling data is analyzed. The logical reasoning method of drilling activity classification using drilling data is established. And the validity of data based on this relationship is verified. Then the original data are processed to form different data sets. This paper builds random forest (RF) and artificial neural network (ANN) models with and without drilling activity classification respectively for a comprehensive comparison. The results show that the ANN model with drilling activity classification outperforms the model without activity classification. For the same data set, the accuracy of the ANN model is improved from 84.71% to 89.58%. Moreover, this model can also achieve less chance of missed warnings. With the comparison of drilling and tripping out that commonly cause the kick, the model accuracy for drilling (85.2%) is lower than that for tripping out (96.7%), due to the influence of more parameters. Among all drilling activities, the warning for drilling inactivity is the best with an accuracy of 99.2%, and the worst one is reaming with an accuracy of 65.9%. For the 33 kick cases, 32 cases are successfully warned using the model. In this study, the activity is recognized first, then the kick is judged by the kick warning model with drilling activity classification. The model can effectively extract the influence of parameter changes caused by the kick and reduce misjudgments. The combined model has strong robustness and good generalization ability, which is expected to facilitate safe drilling.