A New Method of Earlier Kick Assessment Using ANFIS

The late detection of the kick (the entrance of underground fluids into oil wells) leads to oil wellblowouts. It causes human life loss and imposes a great deal of expenses on the petroleum industry.This paper presents the application of adaptive neuro-fuzzy inference system designed for an earlierkick detection using measurable drilling parameters. In order to generate the initial fuzzy inferencesystem, subtractive clustering is utilized. The training set contains 50 data samples and there are 362data samples for testing the proposed method. Also, ANFIS structure is examined at different radii(the parameter of subtractive clustering). Different conformations are tested to get the earliestdetection and the lowest false alarms while facing kick. Eventually, ANFIS verifies the dangerexposure depth of about 28.6 meters before the depth that the kick was sensed by crew. Such anassessment gives the rig crew enough time to prepare for the danger and stop the operation beforebeing exposed to high pressure zones.

Keywords:

Kick , Fuzzy Inference System (FIS) , ANFIS , Subtractive Clustering , Time Series