Adaptation of open-source global optimization software to the process of iterative inversion of resistivity well logs

In the past decades, boreholes were drilled all over the world for the purpose of hydrocarbon prospection. Data from these boreholes are a very valuable resource, that can be used in current geological, geothermal and hydrogeological studies. Since the process of drilling is both expensive and disturbing to the environment the possibility of incorporating data that already exists in the current studies is always worth consideration. However, in the case of older boreholes quality of data is not on par with modern standards which limits its usefulness, especially in the case of data from boreholes drilled in thin-bedded rock formations.

Resistivity logs are one of the main logs used both in hydrocarbon prospection and other applications such as geological, geothermal and hydrogeological studies. Resistivity logs measured by older generations of logging tools are characterized by significantly lower vertical resolution in comparison to logs measured by newer logging tools which affect the quality of the interpretation. However, the information averaged in the process of logging can be partially restored in the process of iterative inversion.

The focus of the presentation is on the utilization of open-source global optimization software in the process of inversion of resistivity well logs. Since inverse problems encountered in geophysics tend to be on the difficult side, relatively simple optimization schemas that often can be found in open-source software are not always giving good results. Therefore, in the presentation, a few methods that allow adapting those algorithms to the problem of inversion of well logs are discussed. The performance of the inversion procedure is validated on synthetic data and real data from the borehole where resistivity logs were measured by different generations of logging tools in the same depth intervals, which allows for comparison of the inversion results to logs measured by modern equipment.

 

The research was funded by the National Science Centre, Poland, grant number 2020/37/N/ ST10/03230.