Estimation of Formation Pressure with Regression Models of Drilling Data

Abstract

Using historical field data from several Louisiana Gulf Coast wells, equations for predicting pore pressure in both normally and abnormally pressured sections of hole were derived by regression analysis. When the pressured sections of hole were derived by regression analysis. When the results of the analysis are applied to drilling data the transition from normal to abnormal pore pressure can be predicted, although only on a geographically regional basis. Introduction Rotary drilling efficiency is closely related to bottomhole cleaning. It is known that drilling with excessive borehole pressure impedes the removal of rock chips, and thereby decreases drilling rate. Adjustment of mud weight to achieve a borehole pressure equal to formation pore pressure significantly pressure equal to formation pore pressure significantly improves the drilling process. Thus, the practice of balanced pressure drilling has become economically attractive. However, because of the inherent danger of abnormally pressured zones and the need for controlling them, numerous techniques have been developed for forewarning. These methods include wireline log interpretation and the analysis of data while drilling. With the latter methods, which require that information on drilling rate, mud, and cuttings be interpreted, there is the advantage of analyzing concurrently with drilling. Interpretation of the many variables logged while drilling is a manual, empirical process consisting of plotting the data as a function of depth and plotting the data as a function of depth and subjectively observing apparent trends. This method is limited by human inaccuracy and personal prejudice. As an alternative to the manual method of interpreting data, a statistical analysis has been performed of information logged while drilling. The numerical method that was used consists of a regression analysis of independent drilling variables and subsequent derivation of a pressure-prediction equation. This equation can be used to estimate formation pore pressure, taking into account varying degrees of pressure, taking into account varying degrees of uncertainty. The purpose of this paper is to discuss the method used to develop the regression equations and show how they may be used to enhance the interpretation process. process. Regression Theory Regression analysis is a statistical method of examining a collection of data to develop a mathematical relationship among variables. The method was used in this study because the properties of the derived equation are statistically defined and the technique is adaptable to high-speed digital computer analysis. Statistical tests on the properties of the equation can determine (1) how well the equation fits the data, (2) the significance of the variables in the equation, (3) the precision of the equation, and (4) the proportion of the variability in the data explained by the proportion of the variability in the data explained by the equation. In developing a regression equation, two types of variables must be defined. The dependent or response variable is the variable to be predicted by the equation, and the independent or descriptive variables are those that affect changes in the response variable. JPT P. 9