| Title | Characterization of Fractures in Geothermal Reservoirs |
|---|---|
| Authors | Egill Juliusson and Roland N. Horne |
| Year | 2010 |
| Conference | World Geothermal Congress |
| Keywords | Fracture, fracture network, deconvolution, well-to-well interaction, reservoir simulation |
| Abstract | It is widely agreed that the flow through geothermal reservoirs is highly fracture-dominated. This has led our research down a path of developing better methods for understanding the fracture networks in the reservoir. In a broad sense, we would like to address questions regarding the extent to which fractures in the subsurface can be characterized. What are some of the more important properties of a fracture network? How much can actually be said about a fracture network given various types of field data? Can we generate realistic numerical simulations of the process? What types of data are most relevant to defining the various characteristics of a fracture network? In an attempt to answer these questions we constructed stochastic fracture network models. Then, simulations with a hydrothermal reservoir simulator (TOUGH2) were used to generate various types of synthetic data. Parts of these data, tracer returns in particular, were further processed to characterize the response in production to tracer injection. The response was characterized by a finite impulse response (kernel) found using a deconvolution technique based on both convex and direct search optimization methods. So far we have shown that a general purpose reservoir simulator can be configured to simulate heat, mass and solute flow through a discrete fracture network. Moreover we have successfully extracted well-to-well transfer functions (kernels) that describe tracer transport through the fracture network quite accurately, given steady state flow conditions. |