It is widely known that shale gas wells require hydraulic fracturing to generate greater wellbore contact with the reservoir in order to achieve economic gas rates. When this type of completion is applied to ultra-tight rock, long-term linear flow is expected and has been widely confirmed through analysis of production data. However, in some shale gas production data, such as the Haynesville, we have observed a markedly different behavior, characterized by the appearance of apparent boundary dominated flow early in the production life of the well. Furthermore, we have observed major inconsistencies between drawdown and buildup data in some of these wells, which is a reliable indicator of degradation in well productivity through time. Although the nature of the productivity loss (kh, skin, fracture conductivity, etc) is usually indeterminable from well performance analysis, its magnitude is clearly evident in well production histories that contain intermittent shut-in periods with measured buildup pressures.
In overpressured reservoirs containing fractures (such as the Haynesville), the reduction of pore pressure resulting from well production through time will sometimes cause the fractures to close, thereby decreasing fracture conductivity. This process, while expected, has not been experimentally verified for the Haynesville (at least within the public literature). We believe that the concept of pressure dependent fracture conductivity (which can be modeled as effective permeability) is a suitable explanation of our observations of well performance data in the Haynesville.
In this paper, we use simulated data to illustrate that the presence of pressure dependent permeability causes a clearly identifiable signature in production history data containing intermittent buildups. In addition, we will show that in the absence of buildup data, analysis of well performance will significantly under-predict original-gas-in-place (OGIP) and expected-ultimate-recovery (EUR) in reservoirs with pressure dependent permeability. Finally, we will show that operating conditions (changing or constant flowing pressure) may have a huge impact on the interpretation of production data, particularly (but not exclusively) in the presence of pressure dependent permeability.