Prediction of Lost Circulation Using Artificial Intelligence in Maroun Oilfield

Nowadays، huge human needs for energy forced petroleum and drilling companies to drill deeper into the earth which means spending more time on drilling and passing through numerous layers with different layer characteristics to hit the target. Since major part of a well cost depends on the duration of drilling phase، an organized drilling program seems vital to save the time and cost. Occurrence of different drilling problems like lost circulation and pipe sticking may deviate the drilling operation from the schedule. Mechanical pipe sticking is likely to occur after complete loss. Lost circulation is one of the common drilling problems in the industry which expose heavy expenses to the oil companies. This problem commences from beginning if the drilling and continues till putting the casing in place. Mud loss can occur in low fluxes up to complete loss that finally can lead to well blowout or severe pipe stuck. Freeing the pipes may waste a week or even more time from the rig. Thus، having accurate information about returned fluid and recording mud loss rate can be great help to prevent drilling problems from taking place. Drilling fluid loss is affected by different factor that make modeling of mud loss suffered from analytical point of view. Thus، employing artificial neural networks which its capability in simulation of complicated phenomena is proven، looks very effective. In this research، using drilling daily report of some wells in Maroum oilfield (SouthWest of Iran)، attempts are directed to predict lost circulation in different areas of this field. Network results in prediction of drilling fluid loss show good compatibility with real data were recorded in drilling daily reports.