New developments in GoFraK to condition naturally fractured reservoir models on dynamic data

@inproceedings{GarciaRM2006,
 abstract = { Modeling of naturally fractured reservoirs generally requires a two-step approach to characterize the fracture systems and convert them into equivalent flow properties for reservoir simulation purposes. In the first step, fracture density models are to be established. They must reproduce well fracturing data but also spatial trends as observed from secondary information related to explicative tectonic or geological episodes. In the second step, equivalent flow properties of fracture systems are to be derived from fracture density maps and other fracture properties. Most critical is the calculation of equivalent permeabilities or more precisely of permeability tensors. The latter are non-linearly related to fracture properties (density, orientation, length, height, and conductivity) and are generally anisotropic with spatially varying anisotropy directions and ratios. Then, and only then, a fractured reservoir model can be flow-evaluated and validated against dynamic data, i.e. well-test or production data. Different evaluation methods can be considered. Whatever the method, mismatches between model’s “flow-responses” and dynamic data are to be corrected by playing with one or several of the many fractured-reservoir model parameters. In practice, the different modeling and calibration steps call for specific tools to account for the discrete nature of fractures, the particular model parameters they involve and the complexity of equivalent permeability tensors. In addition, these tools must be efficient and fast to make reservoir studies both reliable and time-effective. This is the aim of GoFraK, a plugin and workflow in gOcad, which is developed by Total in collaboration with FSS International. The workflow covers the whole modeling chain from fracture-system characterization to model conditioning on dynamic data (see Figure 1). Original and innovative methods have been implemented and already presented for modeling fracture densities and computing equivalent permeability-tensor fields (Figure 2). New developments are now focusing on the way flow responses of GoFraK models can be evaluated and models can be corrected to match better dynamic data. These developments will be presented. They address the following aspects. · Implementation of a flow simulation code taking into account full permeability-tensors. · Flow-based evaluation method of equivalent well-test permeabilities on GoFraK models (Figure 3). · Sensitivity analysis and automatic calibration/inversion methods to make GoFraK models matching dynamic data (Figure 4). },
 author = { Garcia, Michel AND Gouth, François AND Gosselin, Olivier },
 booktitle = { 26th gOcad Meeting },
 month = { "june" },
 publisher = { ASGA },
 title = { New developments in GoFraK to condition naturally fractured reservoir models on dynamic data },
 year = { 2006 }
}