Fractured Reservoir Characterization Using Streamline-Based Inverse Modeling and Artificial Intelligence Tools

Abstract

A new approach that combines the use of static and dynamic data in fracture modeling has been developed. The approach provides the unique opportunity to constrain the fractured models to an effective permeability derived by automatically matching well performances using a streamline-based inverse approach. The effective permeability derived from well performances reflects the stress changes induced by fluid injection, and takes into account the increase of fracture conductivity along the present day maximum horizontal stress. Given the simplicity, robustness, and speed of the streamline based inversion, the estimation of effective permeability is proposed as an alternative to the tedious and often unreliable process of fracture modeling using stress measurements. Because of the availability of a true integrated fracture modeling framework, where any type of data could be used when deriving fracture models, the use of effective permeability derived by inversion is able to reduce the uncertainties by providing models with better fracture prediction capability, as illustrated on a carbonate reservoir. Introduction During the last two decades there have been major advances in the simulation of fractured reservoirs. Most current reservoir simulators are able to simulate dual porosity and dual permeability systems. The new generation of simulators1 are also able to account for the complex geometries created by faults that are commonly found in fractured reservoirs. However, the results of these reservoir simulators are only as good as the input of the reservoir rock properties. To the best of our knowledge, there is yet to be a well defined process for the estimation of 3D fracture permeability, fracture porosity and the shape factor for a dual porosity or dual permeability model. On the other hand, conventional reservoirs have the benefits of geostatistics that provide routinely reliable input for any single porosity model. Furthermore, uncertainties in modeling conventional reservoirs have been considerably reduced by integrating more and more data in the reservoir characterization process. Much effort has been spent on improving the description of the 100 feet surrounding the wellbore where no well will be drilled. Conversely, efforts to find a reliable fractured reservoir model in the interwell region (the one we are really interested in) are very rare. Ouenes et al.2 introduced a collection of artificial intelligence tools to model the interwell region of fractured reservoirs. The details of the method are explained in Ouenes3 and some new developments regarding its combination with DFN models are described by Ouenes and Hartley.4 The method was applied successfully using various static geologic and geophysical data, which are sufficient when the fractured reservoir is under primary production. The use of static data for fractured reservoir modeling is insufficient under secondary and tertiary flooding, which alters the reservoir permeability by increasing the conductivity of the fractures aligned along the present day maximum horizontal stress.5,6 Rather than focusing on stress measurement and stress modeling, we propose a new approach to incorporate the dynamic effects of fluid injection in the fractured reservoir modeling. The new approach uses inverse modeling to estimate the effective permeability, which has been affected by the stress changes due to the injection of fluids into the reservoir.

Application of Artificial Intelligence to Characterize Naturally Fractured Reservoirs

“…al 6,7,8,9,10,11 introduced a collection of artificial Intelligence tools to model the interwell region of fractured reservoirs. al 6,7,8,9,10,11 introduced a collection of artificial Intelligence tools to model the interwell region of fractured reservoirs.…”

Characterization of Heterogeneity Using Inverse Model Equipped with Parallel Genetic Algorithm

“…Moreover, in the case of using GA, forward simulation should be carried out much more, because forward simulation is run for each individual at every generation. Therefore, streamline simulator or neural network simulator have been developed in order to reduce the calculating time 5,6 .…”

Fracture Characterization for Integrated Studies: A New Approach and its Applications

“…Review of the SPE literature reveals that most of the technical papers published on fracture characterization are concluded with the numerical simulation of single well pressure transient analysis 1,2,3 or stream tube modeling between two wells 4 . Few papers mention history matching with 3D flow simulation models as a part of fracture characterization 5,6,7 .…”