Six Years Operating a Real Time Drilling Problem Detection Software in Deepwater Environments: Results and Challenges

Abstract

Abstract Given the complexity and high costs associated to ultra deepwater drilling, any effort to avoid unforeseen situations is beneficial. Initiatives to optimize the drilling process in real-time are extremely important. This article details the experience of 6 year operating a real-time software with the focus of anticipating the detection of potential drilling problems in challenging scenarios. Based on transient numerical simulation (hydraulics, cutting transport, torque and drag, etc.), the software automatically identifies operational problems in real-time, providing to the rig crew anticipated information, so that, corrective and/or preventive actions can be taken. The software incorporate relevant experimental and theoretical research developed in multiple Brazilian universities and R&D Centers. It receives surface and downhole data in real-time, as well as, simulates comprehensive models (which considers real data, fluid properties, well and drilling column geometry as inputs) in order to analyze the current drilling status, by comparing real and calculated (expected) data, with quantitative criteria. Dedicated decision trees, validated by a representative group of senior drillers, provide reliable indication for problem alerts. The present paper aims to detail the successes and the difficulties of detecting typical drilling problems in Campos and Santos basins (offshore Brazil). The process of algorithm development, validation and field implementation is detailed. In July 2014, after almost 10 years of development, the use of the application started as a service conducted by a drilling engineering team, on a 24x7 service in Petrobras Real Time Support Center. Since then, it monitored every Petrobras offshore well. According to estimates, the economy by using this real-time program reached almost 150 rig days, which is equivalent to approximately US$ 130 MM. In 2016, The iniciative was granted the Technological Innovation ANP (Brazilian National Petroleum Agency) Award as recognition by its relevant results. This software is in a continuous development process, aiming to acquire new analyses in order to increasingly help the Petrobras drilling operations. The novelty is the usage of numerical simulation in order to withdraw the subjectivity of a qualitative data analysis, avoiding the misinterpretation of a possible operational event. Present challenges include the proposal of hybrid solutions (modeling + machine learning) to tackle the identification of complex problems (such as false kicks, stuck pipe in salt drilling and circulation loss prediction).