3Norwegian Computing Centre
Fractures can play a significant role in contributing to reservoir permeability in some fields. The chief difficulty is that a sufficiently accurate knowledge of fracture permeability in situ is rarely available. We propose a method of distributing fracture permeability which takes into account test permeability to provide a coherent description of reservoir permeability.
Firstly, an attribute, S, related to fracture permeability is identified and modelled throughout the field. In this case we use a strain model which is considered to provide a good causal link to the fracture permeability. In is important that the variable chosen, S, may be modelled across the field and that at the wells there is a functional relationship between the variable and the permeability. This relationship should be parametric, kf=f(S,Q) with parameters Q. Secondly, we must assign permeability to the fractures. We use an empirical approach. The permeability associated with the fractures is simply the difference between the dynamic permeability obtained from well test and the permeability of the matrix as measured by core, but since kf=f(S,Q) we simply use some curve fitting technique such as regression to find the best Q relating strain to the fracture component of permeability. Since S is modelled over the field, this allows a model of kf in the reservoir. Finally, the well test permeability is a further source of information. The permeability field previously obtained is then modified to account for this extra information. A technique based on kriging was used to account for the test information.
AAPG Search and Discovery Article #90928©1999 AAPG Annual Convention, San Antonio, Texas