80th EAGE Conference and Exhibition 2018: Opportunities Presented by the Energy Transition
In iterative geostatistical seismic inversion methods, the existing experimental data (i.e., the available well-log data) is considered hard-data without any uncertainty. However, some regions of the well may have different levels of reliability, due to, for instance, problems during well-log acquisition. In these cases, the well-log measurements should not be reproduced in the inverted elastic models. Instead, the intrinsic uncertainty associated with each measurement should be included within the inversion procedure since it may carry secondary information that might be useful in guiding the inference of the petro-elastic properties of interest. In this work, we propose a geostatistical framework to account consistently and throughout the entire iterative inversion procedure different layers of uncertainty. In this way, the global stochastic inversion algorithm was adapted to include stochastic sequential simulation with local probability distribution functions as the model perturbation technique to handle local uncertainties along the well path. The proposed methodology was applied to a real case study where we tackled the uncertainty assessment of well-log data and its propagation to the simulation process using stochastic sequential simulation algorithm to incorporate local probability distribution functions associated to the possible acquisition measurement errors due to collapsed zones in the borehole walls. © 2018 Society of Petroleum Engineers. All rights reserved.
Year of publication: 2018