Development and application of regression models for estimation of geological formation pressures

Recently, several oil field service companies have begun offering an expanded mud logging service to drilling contractors and oil companies. A major purpose of this service is to provide early recognition of abnormal formation pressure and estimate the magnitude of this pressure while drilling. Automation of the recognition and estimation of formation pressure is considered in this investigation. Using logging data from six Louisiana Gulf Coast wells pressure prediction equations are developed by regression techniques. Development of decision criteria is discussed and an algorithm for using the prediction equations is presented. A good method is provided for determining the transition from normal to abnormal pressure. Prediction of the magnitude of abnormal pressure is hindered because of the variability in the data. Examination of plots of some of the independent variables versus formation pressure indicates that the regression equations should be developed on a regional basis. A procedure for developing cost equations to be used as decision criteria is presented.