Inversion and uncertainty assessment of ultra-deep azimuthal resistivity logging-while-drilling measurements using particle swarm optimization

As the depth of detection of logging-while-drilling azimuthal resistivity tools increases, more details about formation structure fall into the detection zone. The complicated measurement responses make it impractical to directly interpret the formation from measurements. An inversion process is indispensable to reconstruct the multi-layer formation. Geological inverse problem is known to be non-convex, i.e. the global optimum is surrounded by multiple local optima. Gradient methods are prone to be trapped in local optimum without a good initial model; therefore, in this paper, particle swarm optimization (PSO), a global searching approach, is implemented for the interpretation of ultra-deep azimuthal resistivity measurements. This paper investigates the performance of PSO algorithm through parameter analysis and parallelization study. Also, a gradient method is applied to accelerate the convergence rate. Additionally, to evaluate the risk of solutions, a separate uncertainty assessment procedure is employed. The ability of PSO algorithm is demonstrated by several synthetic examples. Moreover, to justify the capacity of this method, one case study based on field measurements is performed.