Abstract
This article presents an on-going research that addresses the optimization of the cost of drilling wells in environments of high complexity and risk such as those related to the pre-salt region offshore Brazil. The minimization of these costs is directly related to the maximization of ROP (Rate of Penetration). The metric cost, i.e., the cost per meter of drilled not rely solely on the ROP. It directly involves minimization of two major components of that cost: the cost of drills and the operations cost. In short, the better combination number of bits used (generally smaller is better) versus meters drilled, increased ROP and reduced the cost per meter drilled. Finding the best combination is a difficult task. The ideal way to this nirvana is a good planning well and good control of the operation during the process. In such scenarios, it is essential to utilize software tools able to predict and improve the rate of penetration. Such tools must consider both extrinsic drilling skills training, as well as the effects of drilling parameters and drill wear. Such computer systems may, for instance, come to provide reliable alternatives to drilling plans and/or considering alternative plans presented, confirming them or rejecting them based on the information available. The process of creation and implementation of computational models capable of predicting and optimizing the rate of penetration (ROP) in the pre-salt wells is not trivial process. In this research the following techniques are investigated: a Bayesian inference approach for targeting the elicitation process and subsequent combination of models; and a Dynamic Evolving Neural-Fuzzy Inference System (DENFIS). We present the results of this investigation to date, the relevance of the proposed approach and the future prospects of their use for the delivery of viable solutions to the problem.
