Antony Hallam , Debajoy Mukherjee, Romain Chassagne

Well logging is an essential component in the petroleum industry for developing a proper understanding of the subsurface geology and formation conditions.
Unfortunately, the measurements are rarely complete and missing data intervals are common due to operational issues or malfunction of the logging device. Therefore the imputation of missing data from down-hole well logs is a common problem in subsurface workflows.
Recently, many different approaches have been utilised but they are often manual or generalise poorly. Machine learning has reignited interest in this field with promises of a more generic and simpler approach.
We explore whether the chaining of machine learning for mutli-log imputation improves results by overcoming disparities in the patterns of missing data. We will focus this work on the elastic logs of compressional (DT) and shear (DTS) sonic along with the bulk density (RHOB).