An Efficient 1D-2D Adaptive Parametric Inversion Method for Extra-Deep Electromagnetic LWD in Complex Scenarios

Extra-deep electromagnetic (EM) logging-while-drilling (LWD), with its deep investigation and sensitivity to resistivity distribution, is widely used in high-angle and horizontal wells. Due to the complex tool responses of extra-deep EM LWD in heterogeneous formations, eyeball evaluation is often insufficient for accurate bed boundary identification. Therefore, quantitative formation parameter estimation necessitates inversion, with multi-boundary inversion based on 1D models being the most prevalent approach. The inherent limitation of 1D-based inversion is its inability to accurately resolve structurally complex formations, frequently producing oversimplified results. 2D inversion methods alleviate this oversimplification, but their high computational cost limits applicability in real-time geosteering.

In paper, we propose a 1D-2D adaptive parametric inversion framework to combine the efficiency of 1D inversion with the accuracy of 2D inversion. First, we investigate the tool responses and parameter sensitivities in 2D formations using a 2.5D finite-difference algorithm. Then, a scenario-dependent adaptive parametric inversion strategy is developed for specific models based on the sensitivity analysis. For instance, in a fold model, we use a 1D horizontally layered inversion for gently dipping limb regions and a three-point parameterization scheme for the fold core. To improve global optimization, a probabilistic inversion method is established based on the PT-MCMC algorithm, incorporating multiple prior distributions and multiple joint constraints. Finally, the method is applied to the inversion of fold and fault models. Numerical experiments demonstrate that the proposed method accurately reconstructs fold cores and fault structures. Specifically, the relative errors of fault dip angle and fault throw are less than 10%, while the relative error of the fold core location is less than 8%.

Generally, the proposed 1D-2D adaptive parametric inversion framework provides an efficient and robust strategy for real-time geosteering in horizontal and highly deviated wells within complex reservoirs, showing potential for refined reservoir characterization.
We are indebted to the financial support from the National Natural Science Foundation of China (42304140).