Confidence in Subsurface Inversion Models Generated from Electromagnetic Logging While Drilling Data

Identifying a well's stratigraphic position from azimuthal electromagnetic (EM) data requires integrating data from multiple depths of investigation. As a well's position within the stratigraphy can be constantly changing, and formations and fluids show considerable lateral variability, this process is difficult to do manually. To simplify this, inversion algorithms are deployed to represent EM logging while drilling (LWD) measurements as models reflecting the geology. Inversion results are not a direct measurement, therefore confidence in the results is critical. Real-time well placement decisions are routinely made on the output of EM inversions. It is critical to understand that these are models, not direct measurements, therefore verification of the results is essential. This paper discusses the workflows and tools available to interrogate the models generated to give high confidence in the results with a focus on a new deep EM tool deployed in a complex geological environment. The deployment of established EM tools in the same bottom hole assembly (BHA) provides independent verification of the results alongside statistical analysis of the inversion. In many complex depositional environments, the resultant geology is not layer cake. Formations can pinch out or show considerable lateral variability. In these environments it is extremely challenging and sometimes impossible to track a single layer or boundary. We examine a case study from Alaska in a complex shallow marine depositional environment. The target sands were expected to show considerable lateral variability with pinch outs and multiple shale lenses and layers. Deployment of a new, deep azimuthal EM tool with an associated inversion algorithm provided a geological model representing the distribution of the target formations. The stratigraphy was comprised of a complex distribution of sands and shales, many penetrated by the wellbore, with others distributed away from the wellbore based on the depth of investigation of the EM measurements. If this model is the primary tool for mapping the formations and steering to penetrate the most productive zones, it is critical to understand the results and have high confidence in them. The second tool in the BHA, the established azimuthal resistivity tool, provided an opportunity to directly compare the azimuthal data with the inversion result from the new tool to critique the inversion results and help to understand this complex geological environment. The complexity of integrating the data from multiple azimuthal images with different depths of investigation, based on multiple transmitter-receiver spacings and transmission frequencies, demonstrates the need for inversion algorithms to convert the EM field data to a simple-to-understand representation of the geology. This case study provides proof of the quality of the model, especially in such a complex geological environment, allowing high confidence in the deployment of this new tool for well path optimization.