1-d Inversion of Triaxial Induction Logging in Layered Anisotropic Formation

In this paper, we present a one-dimensional (1-D) inversion algorithm for triaxial induction logging tools in multi-layered transverse isotropic (TI) formation. A non-linear least-square model based on Gauss-Newton algorithm is used in the inversion. Cholesky factorization is implemented to improve the stability and the reliability of the inversion. Zero-D inversion is conducted at the center of each layer to provide a reasonable initial guess for the best efficiency of the inversion procedure. Cross components are used to provide sufficient information for determining the boundaries in the initial guess. It will be illustrated that using all the nine components of the conductivity/resistivity yield more reliable inversion results and even faster convergence than using only the diagonal components. The resultant algorithm can be used to obtain various geophysical parameters such as layer boundaries, horizontal and vertical resistivity, dipping angle and rotation angle etc. from triaxial logging data automatically without any priori information. Several synthetic examples are presented to demonstrate the capability and reliability of the inversion algorithm.