ABSTRACT: Application of neural net technology and climate-stratigraphy in stratigraphic correlations

Fonseca, C., M. Schaaf, and C. J. van der Zwan , Shell International Exploration and Production BV, Rijswijk, Netherlands

In brown field areas reservoir units need to be correlated at a resolution far beyond that can be achieved by seismic or by other stratigraphic techniques. The stratigraphic framework is normally based on wireline log correlation, controlled by various types of qualitative bio and non-biostratigraphic dating tools. However, all these tools have limited resolution. Higher resolution is nowadays attempted by applying quantitative dating techniques with limited effect. Limitations of these techniques are due to the constraints of the stratigrapher to evaluate all available information objectively.

We have combined neural network technology and climate signals in sediments to create high-resolution stratigraphic correlations that make use of meaningful quantitative information such as well logs, bio- and non-biostratigraphic data, and cyclostratigraphy. Datasets from terrestrial Carboniferous sediments and from Pliocene deep-water sands were used for the evaluation of this innovative technology. The results achieved with the neural net represent a 5 times improvement in resolution (up to 100 ky) and precision of stratigraphic control in reservoir correlation. The coupled application of neural net technology and climate stratigraphy increases the reliability of the stratigraphical framework in the sub-surface while providing a better definition of reservoir distribution to obtain better reserve estimates and increases in ultimate recovery.

AAPG Search and Discovery Article #90913©2000 AAPG International Conference and Exhibition, Bali, Indonesia