It is standard industry practice to use three-dimensional geological models to predict rock properties of a reservoir.
These ‘static’ geological models contain information at high resolution about the distribution of rock types (facies) usually interpolated from core analyses, and wireline information, measured at discrete well locations often kilometres apart. Additional parameters such as porosity and permeability are distributed throughout the model according to the spatial variability typically associated with the interpolated facies or from inversion of seismic information.
An alternative method, stratigraphic forward modelling (Sedsim), numerically simulates depositional processes such as erosion, transport, deposition, and compaction to predict lateral and vertical variations in reservoir and seal properties associated with predicted lithofacies distribution. An advantage is that forward numerical models may be constructed using a limited amount of data, useful for greenfield sites, and the resulting models may also reduce depositional uncertainty for the same amount of input data required by more conventional static methods.
Because of the initially limited well and seismic data in the South West Hub area, ANLEC R&D identified the need to study the utility of forward stratigraphic modelling and sponsored a two phase project to:
- construct a static, cellular geological model for the area using Sedsim stratigraphic forward modelling software; and
- use this model as the basis for dynamic flow simulation and contrast the results with those obtained using a conventionally derived static model.
Phase 1 used Sedsim to create a stratigraphic forward model from 250 Ma to 182 Ma (Triassic to Lower Jurassic), from the base Wonnerup equivalent to top Eneabba equivalent, that predicted the distribution of grainsize and primary porosity (and permeability via a transfer function) of sediments below seismic resolution around the Harvey-1 well.
Phase 2 used this model as a basis to perform a flow simulation of CO2 into the Wonnerup Sandstones using a nested 500m grid and dynamic model area of 25 x 15km. The simulation of CO2 injection used Eclipse 300 software and the results were compared to a previous simulation performed by Schlumberger based on a static model constructed using well data and geostatistical distribution of reservoir properties.
Both methods predict CO2 plumes to develop and remain around the injection wells for the long term; the main difference being that the plume predicted by the ‘Sedsim’ input spreads more than the conventional model.
This was the first time such a comparison of dynamic simulation of geological models generated by these different methods has been performed and has provided insights into dynamic models for CO2 sequestration.
Having an alternative approach to depict the geological framework allows building confidence and reducing risk and cost of obtaining accurate representation of the reservoirs and the forward models.