Improving Drilling Efficiency and Reducing Well Construction Period Based on Rig State Classification Using Machine Learning Method

In the informatization and intellectualization era of oil and gas, automation is essential to measure and track detailed performance for routine drilling operations by automatically measuring these individual operations consistently. Bohai Oilfield is the largest offshore oilfield in China with annual production of 30 million tons. More than 400 wells are being drilled there every year to maintain this production, an intelligent decision system for drilling has been established to reduce non production time and invisible lost time of these wells. The data are collected and transferred to the remote support site using Internet of Things from the sensors on the rig site. Based on this, a rig state classifier using Long Short-Term Memory combined with Whale Optimization Algorithm is established. The accuracy of WOA-LSTM model in classifying 9 routine rig states using 7 comprehensive logging parameters is up to 96%. This is one of the first attempts for improving drilling efficiency considering comprehensive data using machine learning. This methodology has been successfully applied to drilling performance improvement in drilling campaign of Bohai Oilfield, and it can also be a reference to other oilfields.