Real-time prediction method of borehole stability

Abstract

Based on the close relationship between seismic and logging information, a real-time prediction model of borehole stability is established using seismic, logging, and geological data to control borehole wall sloughing instability. First, seismic attributes are extracted from borehole-side seismic traces of target wells and drilled offset wells. The mapping models of relationships between seismic attributes and logging data of various formation intervals in drilled wells are then constructed using wavelet neural network. Using the seismic attributes of formation under bit and the corresponding mapping model, the acoustic and density logging data of the current undrilled formation can be predicted. On the basis of the prediction results, the mechanical model of borehole stability is employed to calculate pore pressure, collapse pressure, and fracture pressure, thus predicting the safe drilling fluid density range. Practical application in Tarim Oilfield shows that real-time operation performance of the model is excellent and the prediction accuracy of parameters is satisfactory. Key words: borehole stability, real-time prediction, seismic attribute, wavelet neural network, safe drilling fluid density