Service Communautaire d'Information sur la Recherche et le Développement - CORDIS

dGB-GDI software portfolio

dGB-GDI is a general purpose, highly flexible, quantitative interpretation software system for seismic lateral prediction, seismic patern analysis and rockphysical/petrophysical analysis. The system consists of modules which are organised according to functionality into 6 menus.
At the core of the system is the integration framework which specifies what data is to be studied and how the data should be integrated. In a fully integrated dataset stratigraphy is linked to well logs, to engineering data and to seismic measurements. The more advanced techniques can only be carried out if the data have been integrated first. However, it is possible to use the dGB-GDI tools without integrating the data.
Supervised learning techniques can be applied without fully integrating the data first. In existing fields neural networks can be trained on real data. Networks can be trained to classify or quantify the seismic measurements. To apply this technique only the seismic data, the interpretation horizon and the essential well data must be loaded.
If only a limited number of wells is available to train the neural network, it is better to extend the training set with stochastically simulated wells. However, as with analysis of unsupervised seismic patterns, for optimal simulations a fully integrated dataset must exist.

Stratigraphic sequences can be analysed using generalised Markov chains. Wells with different sedimentary stacking patterns are easily detected by this method. The result of the sequence analysis can be used by the pseudo-well simulator to generate wells with similar stacking patterns.
Another advanced inversion technique is the Model Probability Module (MPM). In MPM simulated pseudo-wells are scored at real seismic locations for seismic response and geostatistical probability, which yields a 'probability' for each analysed location. Scores are subsequently analysed to yield expectation and standard deviation grids per quantity.