Predicting sonic logs from petrophysical logs in oil and gas wells

What is it about?

Multivariate linear regression is an important statistical tool to find relationships between different measurements. We use this together with interpretation of geology to show, that sonic logs can be estimated and thus predicted reliably from other petrophysical logs commonly measured in oil and gas wells.

Featured Image

Why is it important?

Many wells in producing oil and gas fields do not have sonic logs recorded, often because of mechanical issues with deviated wellbores or high cost. Sonic logs are particular important to tie 3D seismic data to the subsurface geology. If sonic logs measurements can be predicted, then we can also predict how the changing geology of reservoirs might look like on 3D seismic data, and how the changes in the fluid fills of the reservoirs will look like on 3D seismic data, like replacement of oil with water, or oil with gas. The statistical tool not only gives a prediction but also measures the reliability. This makes the sonic log prediction an important tool in reservoir development and management.