Abstract

Formation pore pressure is the pressure of fluid in the formation pore, which is an important parameter in oil and gas well engineering. It is related to the design of the well structure, drilling fluid density, and the supporting use of casing pipe. Accurate prediction of formation pore pressure contributes to achieving safe drilling. However, due to the complex geology of carbonate rock formation, it is difficult to predict. This paper presents a prediction method for carbonate formation pore pressure based on machine learning. This method first analyzes the causes of abnormal formation pore pressure in the Block X and selects the Fan comprehensive interpretation method and the grey correlation degree method as the theoretical basis to determine the correlation between formation pore pressure and different well-logs. Through the Back propagation neural network method, the relationship between formation pore pressure and well-logs is established, and the prediction results have higher accuracy than the original model. Compared with the measured data, the error is less than 5%, which meets the needs of engineering. Compared with traditional methods, this method has strong applicability and greatly improves the accuracy of prediction, which is suitable for the pressure prediction of multiple genesis in carbonate rock formation.

This content is only available via PDF.
You can access this article if you purchase or spend a download.