Data for: Learning about risk: machine learning for risk assessment
In order to avoid potential damage during drilling operations for a new offshore Oil&Gas well, a semi-submersible drilling unit should maintain the position above the wellhead. This is particularly critical if the platform is located in shallow waters, where small changes of position lead to higher riser (pipe connecting the platform to the subsea drilling system) angles. Exceeding physical inclination limits may result in damages to wellhead, Blowout Preventer (BOP – sealing the well) or Lower Marine Riser Package (LMRP – connecting riser and BOP). Platform positioning is maintained in an autonomous way through the action of a set of thrusters controlled by the Dynamic Positioning (DP) system. Platform position may be lost due to a series of reasons. In this case study, it is assumed that the platform thrusters exercise propulsion towards a wrong direction, leading to a scenario of "drive-off. If the rig moves to an offset position, specific alarms turn on and suggest the DPO to stop the drive-off scenario by deactivating the thrusters and initiating the manual Emergency Disconnect Sequence (EDS) for the disconnection of the riser from the BOP. If the manual EDS ultimately fails, the automatic EDS activates at the ultimate position limit allowing for safe disconnection. Matteini (2015) studied in detail occurrence and development of drive-off scenarios. Relevant indicators were defined to assess the performance of systems and safety barriers. Indicator trends were simulated for a period of 30 years. The database includes all the simulated indicator values transformed into their derivative with respect to time t as inputs xi to the machine learning model. Since the simulated wellhead damage frequency Freq is an expression of the scenario probability p, and, in turn, the risk R, for constant scenario s and consequence c, we can state that: dFreq/dt≈dR/dt After Freq is also transformed into its derivative with respect to time t, a label y indicating frequency/risk increase or decrease is added to the database. Two datasets created from the overall database can be also found: 1. Training dataset used to train the DNN model, with 2/3 of the xi and associated y values (160), and 2. Test dataset used to test the DNN model, with about 1/3 of the xi and associated y values (79).