The rapid pace of capital investment in the Haynesville Shale in North Louisiana and Texas from delineation in 2008 to full-field development in 2010 has focused the need to develop a work flow which leads to identifying " sweet spots" of production to optimize the return on capital. 3D seismic has already proven beneficial in hazard avoidance while drilling the Haynesville and other unconventional plays such as the Barnett shale. However, advances in seismic acquisition and processing coupled with pilot hole and core data have led to a more predictive use of 3D data. It is recognized that all unconventional shale plays have vertical and lateral heterogeneity of reservoir storage, " fracability" and hydraulic fracture containment. We show results from a 300 mi2 study area in North Louisiana comprised of a contiguous, multi-client 3D seismic survey acquired from 2008 to 2011. The survey was designed to image the Haynesville target from 11,000 ft. to 14,000 ft. with a wide-azimuth and long offset design. Eight vertical pilot holes with full-suite logging are located within the study area, allowing for tight constraint on the post-stack inversion and estimate of rock properties away from well control.
Model based inversion was used to generate a P-Impedance volume. Probabilistic neural networks (PNN) were then used to estimate rock property volumes which were guided by pilot logs, the P-Impedance volume, and the input seismic volume. Target rock properties that were considered most likely to yield results were Density, Poisson's ratio, Young's modulus, Mu-Rho, and Lambda-Rho. Overall, attributes from the Haynesville interval were extracted from 17 seismic rock property volumes and averaged along each horizontal well path to arrive at one value per attribute per well. We illustrate the change of Haynesville reservoir properties trending across the study area.
With several years of gas production now available in the Haynesville from hundreds of horizontal wells, Estimated Ultimate Recovery (EUR) projections have been generally de-risked and were cross-plotted against reservoir rock property estimates from seismic. Because well productivity is sensitive to varying completion parameters, a critical step is inclusion of completion metrics to minimize the variability due to the completion. We found that an EUR model which included the effect of completion parameters combined with the Mu-Rho rock property (" fracability") from seismic resulted in the best correlation to well performance.