Study On The Methods Of Formation Characteristic Parameters Prediction In Drilling Simulation

During the oil&gas prospecting and exploitation, formation characteristic parameters prediction is an important study, it can help to ensure safety in drilling, raise the drilling efficiency and reduce the drilling costs, et al. Therefore, the study of the methods of formation characteristic parameters prediction is of momentous practical significance.Based on the key project of SinoPec---Drilling Simulation based on Geological Database of Drilling Engineering, this thesis mainly studies on the methods of formation characteristic parameters prediction in drilling simulation. Its main contributions are as follows:1. Lithologic identification based on well logging dataBy analyzing the correlation between well logging data and lithologic information, the radial basis function (RBF) neural network is used to found mapping model from well logging data to lithology. Experimental results show that the proposed method is simple, fast and suitable for lithologic identification.2. Pore pressure and fracture pressure evaluationIn this thesis, support vector machines (SVM) is used to establish pore pressure prediction model based on the effective pressure theorem and the sonic velocity model. Practical application shows that result can be obtained with this model for the shale sand formation. Compared with the traditional pore pressure estimation methods, this method doesn't need any normal compaction trend with good adaptability and accuracy.3. Estimation of the formation drillability parameterIn this thesis, we also present a practical approach to calculate the formation drillabiltiy parameter from well logging data based on BP neural network. After analyzing the relations between the log information and the formation drillability, some well log parameters related to formation drillability are selected, and a mathematical model for predicting the formation drillability is established by BP neural network. Experimental results show that this method can predict formation drillabilty parameter satisfactorily.