Parameters Prediction In Geological Environment Modeling For Drilling Engineering

The correct lithologic identification and the accurate prediction of formation pressure are the base for the optimization of bit type and drilling parameter, the proper determination of mud density and the optimization of well trajectory; they are also the premise of the reservoir protection, the improvement of drilling success rate and the reduction of drilling cost.Based on the key project of SinoPec---'Drilling Simulation based on Geological Database of Drilling Engineering', this thesis mainly studies on the method of formation characteristic parameters prediction in geological environment modeling for drilling engineering. Its main contributions are as follows:1. Novel natural gamma ray prediction method based on support vector machineBy analyzing the mapping relation between the seismic data and well logging information, a novel natural gamma ray (GR) prediction method was presented. The prediction model for GR was established with support vector machine (SVM) before drilling by using the seismic and well logging data. The proposed method was applied to predict GR of the well Yong in Junggar Basin. The experimental results show that the prediction effect of SVM method is better than BP method based.2. Formation pressure predictionA pore pressure prediction model was established based on effective pressure theorem. This process was implemented with support vector machine. The method using sonic interval transit time regression to determine the effective stress and the one using sonic velocity model were compared. The results show that it is feasible to predict pore pressure by using effective pressure theorem; the method using sonic velocity model should be taken into consideration firstly.