Reservoir production can be stimulated by creating hydraulic fractures that effectively facilitate the inflow of hydrocarbons into a well. Considering the effectiveness and safety of the operation, it is desirable to monitor the size and location of the fracture. In this paper we investigate the possibilities of using seismic waves generated by active sources to characterize the fractures.
First, we must understand the scattering of seismic waves by hydraulic fractures. For that purpose we use a finite-difference modeling scheme. We argue that a mechanically open hydraulic fracture can be represented by a thin, fluid-filled layer. The width or aperture of the fracture is often small compared to the seismic wavelength, which forces us to use a very fine grid spacing to define the fracture. Based on equidistant grids, this results in a large number of grid points and hence computationally expensive problems. We show that this problem can be overcome by allowing for a variation in grid spacing in the finite-difference scheme to accommodate the large-scale variation in such a model.
Second, we show ultrasonic data of small-scale hydraulic fracture experiments in the laboratory. At first sight it is difficult to unravel the interpretation of the various events measured. We use the results of the finite-difference modeling to postulate various possible events that might be present in the data. By comparing the calculated arrival times of these events with the laboratory and finite-difference data, we are able to propose a plausible explanation of the set of scattering events. Based on the laboratory data, we conclude that active seismic sources can potentially be used to determine fracture size and location in the field. The modeling example of fracture scattering illustrates the benefit of the finite-difference technique with a variation in grid spacing for comparing numerical and physical experiments.