A new method for detecting real-time geopressure from drilling-logging parameters

Abstract

Real-Time detecting abnormal formation pressure can not only prevent the happening of drilling hazard, but also effective protect the pollution of reservoir. A detective model can be made from some drilling-logging parameters because these parameters collected by comprehensive logging instrument can indicate the abnormal pressure information existing in the formation. First, a PCA method is used to process six wells from Dongying Depression, China in order to reduce the cross-correlation among parameters and the count. Then a neural net model is trained by the result in the first step. Finally, thirty wells are detected by the model. The correspondence between real data and predicted results is about 84.6%.So this method can be used in the real case.

Multi-Parameters Predicting Method of Geo-Pressure in Carbonate Formation

“…Because Principal Component Analysis (PCA) has advantages of moisture resistance, energy conservation, de-correlation and the redistribution and concentration of energy, it is an optimal orthogonal transforms in mean variance principles and a multiple statistics analysis which can choose important variables by linear transformation. PCA can also create less new uncorrelated variable from the entire variable proposed before which can keep all its original information in reflecting formation pressure [8,9].…”