Global Sensitivity Analysis of Array Laterolog Response in Anisotropic Formation Based on Morris Method

Abstract High-angle and horizontal wells are used in unconventional hydrocarbon exploration and development widely. The influence of formation resistivity anisotropy on logging measurement is significant in deviated and horizontal wells. Affected by mud invasion, the influencing factors of array laterolog apparent resistivity response are multiple and complex. Those factors affect each other with nonlinear characteristics, which causes ill-conditioned problem in the inversion processing of formation parameters. Therefore, it’s necessary to analyze the sensitivity of various factors and extract the primary controlling factors of logging response. The forward modeling of array laterolog response is established based on FEM (finite element method) in deviated and horizontal wells with anisotropic formation. The potential field distribution of different detection modes in isotropic and anisotropic formations is compared. The interactive effect between the anisotropy coefficient and the well deviation angle, the invasion depth, the formation thickness on the array latrrolog response is simulated respectively. Using Morris global sensitivity analysis method, the sensitivity and interaction degree of various influencing factors are simulated. The result indicates that the anisotropy has a significant impact on the apparent resistivity when the deviation angle is greater than 40°. Mud invasion in anisotropic formation can cause an increase in apparent resistivity and curves disorder. Anisotropy in thick layer causes apparent resistivity curves disorder and increase the degree of curve separation. Mode 1 and mode 2 with shallow detection depth, the response is most sensitive to the change of borehole radius. Mode 3-5 with large detection depth, which response is most sensitive to the change of the horizontal resistivity of uninvaded formation. The layer thickness and well deviation angle show a significant interaction with other factors. The results provide a basis for parameter decoupling in the array laterolog logging data hierarchical inversion processing