Reducing Uncertainty in Exploration and Development by Incorporating Targeted Machine Learning Interpretation Workflows in Jurassic Carbonate Reservoirs

Abstract

Presents a case study from the Middle East giant fields in which uncertainty is reduced in the exploration of carbonate stratigraphic traps. Machine Learning interpretation workflows used in the evaluation involve unsupervised seismic facies interpretations for gross depositional mapping of a stacked reservoir system, along with well log unsupervised facies classification, to define facies and heterogeneity of the reservoir units. These workflows are integrated, along with semi-automatic seismic interpretation and fluid substitution techniques, in a study area with several exploration wells. The fluid substitution techniques involve use of petrophysical and elastic rock properties crossplots, and fluid characterization, to delineate reservoir intervals which have a lower risk for hydrocarbon accumulation. This topic is relevant to the workshop owing to risks associated with carbonate stratigraphic trap identification and delineation, petroleum system challenges and reservoir heterogeneity issues identified in exploration and development wells in the region.

AAPG Datapages/Search and Discovery Article #90361 © 2019 AAPG Latin America & Caribbean Region Geosciences Technology Workshop, Recent Discoveries and Exploration and Development Opportunities in the Guiana Basin, Paramaribo, Suriname, November 6-7, 2019