Increasing the predictive power of geostatistical reservoir models by integration of geological constraints from stratigraphic forward modeling

• Stratigraphic Forward Model (SFM) provides channel body volumetrics at basin scale.
• Basin-scale constraints from the SFM were integrated into reservoir models.
• Uncertainty of basin-fill parameters was propagated to reserve estimation.
• Inference of basin-fill parameters from reservoir data reduced their uncertainty.

Current static reservoir models are created by quantitative integration of interpreted well and seismic data through geostatistical tools. In these models, equiprobable realizations of structural settings and property distributions can be generated by stochastic simulation techniques. The integration of regional (or basin) scale knowledge in reservoir models is typically performed qualitatively or semi-quantitatively (for example, through the definition of regional property trends or main channel-belt orientations). This limited use of regional information does not allow an assessment of the impact of the uncertainties associated with the regional knowledge on the overall uncertainty of the reservoir model.A novel approach is proposed in this study, which allows us to consistently integrate basin-scale information into reservoir models. A new type of data, related to the distribution of the potential hydrocarbon-bearing volumes at basin scale, was obtained from a 2-DH process-based stratigraphic forward model (SFM) and integrated as a soft constraint in the geostatistical reservoir modeling. As a consequence, reservoir models are quantitatively consistent with the large-scale geological setting defined by the SFM output. Furthermore, the uncertainty associated with each SFM parameter can be propagated to reserve estimation. Thus the partitioning of the overall uncertainty affecting a reservoir model into the contributions of the uncertainties at the basin and reservoir scales can be quantitatively assessed.Several synthetic case studies were carried out with and without conditioning to SFM output, which verified the effectiveness of the method. A logical next step is to apply the proposed methodology to a real-world case.