--> Integrated Static-Dynamic Reservoir Modeling of a Deep-Water West African Reservoir Utilizing an Efficient Decision-Based Workflow

AAPG ACE 2018

Datapages, Inc.Print this page

Integrated Static-Dynamic Reservoir Modeling of a Deep-Water West African Reservoir Utilizing an Efficient Decision-Based Workflow

Abstract

This case study demonstrates the use of a multi-disciplinary decision-based methodology to evaluate geologic uncertainty in a highly heterogeneous reservoir. The reservoir consists of levee confined channels characterized by core-calibrated log data, high-resolution 3D seismic data, analogs and well tests. The field has been producing oil for a year with support from water injection.

In the framing phase, key uncertainties were identified and combined in a decision tree to define multiple scenarios with various levels of complexities. Elements affecting flow and their implementation in the models were considered such as channel stacking pattern and geometry, distribution of sandy petrofacies and baffles/barriers, and variability of porosity/permeability. Models were built over a small but representative area in order to accelerate learnings from dynamic simulation and iterative updates of the seismic interpretation and static models.

Models with detailed seismic stratigraphic mapping, layers honoring the channel axis-margin-levee geometries, and multiple facies having distinct porosity/permeability trends showed better history matching results than more simplistic models. These findings narrowed the range of uncertainty by constraining the methodology for mapping channels and representing subseismic baffles and barriers. Second order uncertainties were represented by equally likely models, used as input in an integrated static-dynamic uncertainty workflow. The study defined an efficient workflow to create low-mid-high case models for the full field that are not anchored to a base case. This was achieved in a limited amount of time with the geologist, geophysicist, petrophysicist, and reservoir engineer working concurrently on a limited set of data rather than sequentially over the full field. We believe this represents a more efficient, decision-based alternative to the standard interpretation and modelling workflow.