Generating Missing Logs

Overview

We have two methods for generating pseudo-logs from other logs:

We can use our Esteem module of LESA to generate a sonic using a neural network.
We can also use special deterministic approaches in the Synthetica module of LESAfor generating a sonic from a VSH, Density or Density/Neutron combination.

These two techniques can be used independently or in conjunction to generate missing sonic logs.

Differences between the two pseudo logs can be used to define changes in rock lithology and to pick formation boundaries.

Examples

Application of an Esteem neural network to create a pseudo sonic log using a gamma ray and deep resistivity only.

The training was done on a well 2 miles away.

Example shows excellent agreement with actual sonic. Note: the neural network did not "know" this sonic exists.

Example of a Cretaceous gas well from Wyoming shows comparisons between an Esteem neural network pseudo sonic and a deterministic theoretical sonic.

We see excellent comparisons when we plot the difference between the neural network pseudo sonic and the actual sonic, versus the difference between the deterministic theoretical sonic and the actual sonic.

This indicates the two very different techniques are converging on the same answer.