What GAMLS Does

GAMLS - Geologic Analysis via Maximum Likelihood System

What GAMLS Does

GAMLS does the following:

1. Analyzes digitized wireline well log data, plus core data if it exists.
2. Clusters datasets into modes (= clusters = rock types = flow units) using a model-based, maximum likelihood system wherein each sample is assigned to each mode with a fractional probability (mode probability assignment). Therefore, samples (depths) assigned to the same cluster have similar log properties.
3. Permits correlation of rock units between wells.
4. Generates graphics (plots) and tables that show the relationships among variables for each mode.
5. Permits predictions of missing data using the statistical relationships computed during a previous clustering run.

Clustering

Clustering is the grouping together of rocks (digitized depths) that have similar log signatures. The number of groups (clusters, rock types, flow units ....) is determined and the log tools to use is determined by the user. In this way, the user has much control on the output. (click for more on clustering)

Correlation

Correlation (between wells) is done by clustering two or more wells within the same computer run so that like modes identified in each well have similar log characteristics. Then, the beds for each mode are correlated in the usual manner.

Prediction

Prediction is accomplished by relating the log tool signals to known values of some characteristic of the rock. Examples of variables that can be predicted are permeability, cementation exponent, a missing log tool, etc. In order for prediction to be possible, there must be enough empirical data (e.g., laboratory measured core permeability values) so that the n-dimensional correlation between the log tools and the emiprical values can be made by GAMLS.

GAMLS is particularly powerful in that it:

1. requires little training relative to non-model based neural systems; and
2. is extremely flexible in its ability to use databases with much missing data (without the need to interpolate, extrapolate, or estimate the values of the missing data.)

Return to Eric Geoscience Inc. Home Page