| Popis: |
Well planning is one of the most important steps of drilling engineering to avoid problems such as kicks, well instability, blowouts, and lost circulation, which represents the main causes of environmental damage, economic loss, and human loss included. These damages are especially critical when it comes to ultra-deepwater drilling. Usually, to estimate pore pressure, methods based on historical experience, mathematical statistical, empirical, and field-measured are used. However, accuracy on each method is a challenge for improvement, as they depend on how the instrument or petrophysics analysts reads the particular measurement. Therefore, methods such as artificial neural network (ANN) are studied, aiming to find values of formation pressures with lesser margin of error. Besides being a faster and more accurate method than the conventional ones, the ANN is an adaptive computer system inspired by the human brain itself, that acquires knowledge with experience. The outline is to use this artificial intelligence to estimate pore pressure. Therewith, a case study was performed to create, train, and implement ANN from data logs of the well studied, besides depth and density of rock formation information. The proposal is to estimate pore pressure using this artificial intelligence technique and compare the obtained results between the Eaton conventional method and the created ANN, in order to prove that the results using this new technique reach smaller errors. |
