Inversion of Three-Dimensional Exploration-Scale Stratigraphic Forward Models for Derisking Reservoir Presence: Application to the Tertiary GoM

Conventional hydrocarbon exploration is increasingly challenged, with a shift towards deeper, more difficult plays. Within this context, understanding and predicting reservoir distribution at regional to basin scales is a key element of risk polarization. One method to assess this risk is stratigraphic forward modelling. Traditionally, however, such models have been constrained by a manual calibration approach. To move beyond the limitations of such methodology, we use a stochastic inversion algorithm coupled to the DIONISOS 3D stratigraphic forward modelling package. DIONISOS can simulate erosion, transport and deposition of sediments at basin scale and over geological time.

The inversion algorithm, which is a modification of the so-called “neighborhood algorithm”, combines multiple error functions with user-defined error thresholds, and aims to find acceptable models that reproduce available data constraints within certain error ranges, while accounting for uncertainty in the forward model assumptions. The process is accelerated by parallelization using computer clusters. Probabilistic cluster analysis is applied to the accepted models in order to identify discrete sets of geological scenarios and understand their relation to underlying data and forward model predictions. Finally, these geological scenarios are used to predict broad reservoir presence in areas devoid of data constraints.

The inversion methodology is illustrated on a case study for deepwater exploration in the GOM. The models cover a 400 x 460km area, from the delta to the abyssal plain, with a grid cell size of 10 x 10km, and simulate almost 6 my of geological time. More than 70000 models were evaluated, from which 317 were obtained that agree with available well data and prior geological knowledge, within the error ranges predefined. The cluster analysis of these models reveals 9 different classes of scenarios with distinct geologic input parameters that lead to distinct reservoir presence predictions away from available control data.
 

AAPG Search and Discovery Article #90142 © 2012 AAPG Annual Convention and Exhibition, April 22-25, 2012, Long Beach, California