Publication | Closed Access
Fast and Efficient Modeling and Conditioning of Naturally FracturedReservoir Models Using Static and Dynamic Data
13
Citations
0
References
2007
Year
Unknown Venue
A large proportion of petroleum reservoirs is known to be naturally fractured with consequences on their flow behavior hence on reservoir performance. Though the modeling of such reservoirs has been the purpose of many research works, it remains a challenging task. Too simplistic reservoir models do not allow capturing essential features like large-scale fracturing trends, or non-linear multivariate relationships between the equivalent (generally anisotropic) permeability of the fracture system, and fracture densities and properties to be characterized on a directional fracture-set basis. Conversely, too complex reservoir models, intended to be more realistic, require computationally intensive and memory consuming algorithms. They also involve numerous parameters, a large part of which cannot be estimated from available data.In-between, there is a need for reasonably complex models and methods to generate them in a consistent way with various fracturing and dynamic data in order to produce conditional models. This paper presents such an approach, which has been developed as a workflow.The approach is based on an original conceptual model of fracture systems and a notion of scale-dependent effective properties. It is also a two-step modeling approach in which the fracture system is first characterized, then converted into equivalent flow properties for reservoir simulation purposes. Key aspects of the approach include the geostatistical modeling of fracture densities, scale-dependent calculation of equivalent within-layer horizontal permeability tensors based on spatially periodic discrete fracture networks, analytical calculations of vertical inter-layer permeabilities, and conditioning to well-test permeabilities by using steady-state flow-based evaluation of reservoir model responses. All these aspects rely on innovative and CPU-time efficient methods. They are introduced and illustrated by case-study results.