Estimation of 2D and 3D In-situ Stresses Using Back Analysis of Measurements of Well / Borehole Deformation
MetadataShow full item record
In-situ stress state in underground plays a significant role in planning, design, ground support, well drilling and wellbore stability control. The overcoring method with strain gauges on borehole wall is the only method for determining the complete 3D stresses. However, this method requires physical access and cannot be applied to petroleum engineering. The purpose of this research is to develop a practical back-analysis method using measured borehole deformation to determine a) the 2D in-situ stresses in the plane perpendicular to the borehole as an alternative method, and b) most importantly the complete 3D in-situ stresses for use in petroleum and other rock engineering. The difficulty in determining the 3D stresses from 2D borehole deformation measurement is overcome with differential-direction drilling. This requires diametrical convergence measurement in three non-parallel planes as a minimum. For petroleum application, this can be achieved by measurements in three different sections of a directional well with a total inclination angle interval between the measurement planes at least 35°. The three sets of measured convergence are combined through comprehensive coordinate and direction relationships. Conditions in petroleum fields are also considered and a total of five comprehensive models are developed based on permeability, pore and mud pressure in both 2D and 3D analyses. In addition, statistical approaches are applied to help detect erroneous data and search for the best-fit solution. To facilitate calculation, automated analysis packages based on excel have been developed for different scenarios in both 2D and 3D analyses. A standalone software for 2D analysis is also developed. The developed method has been verified using simulated measurement data. The results show that the back-analyzed stresses agree well with the applied stresses when input errors are small. Even with up to 30% input errors, the result converges to the real solution with smaller difference between the back-analyzed stresses and the real solution. This demonstrates the accuracy and reliability of the solutions, and confirms the validity of the developed method and the analysis procedure.