NUMERICAL INVESTIGATION OF METHANE AND FORMATION FLUID LEAKAGE ALONG THE CASING OF A DECOMMISSIONED SHALE GAS WELL
A conceptual model of a vertical decommissioned shale gas well is first developed, based on a geological sequence in the St. Lawrence Lowlands Basin (Quebec, Canada), containing from bottom to top, 200 m of Utica shale overlain by 750 m of low-permeable caprock (Lorraine Group), which is in turn overlain by a 50 m thick surficial sand aquifer. The geological sequence is intersected by a borehole with an external diameter of 200 mm and a casing annulus thickness of 50 mm. A parametric study of leakage scenarios is then conducted, assuming isothermal multi-phase, multi-component flow. The model is not calibrated, but the simulations are used to identify the possible range of leakage rates and the time scale for gas and contaminant migration along the casing annulus. Although observed properties of the geological formations of the St. Lawrence Lowlands Basin are used, the results presented here have broad outreach due to the similarity of hydrodynamic properties of other shale formations.
The numerical results show that an adequately cemented borehole (with a casing annulus permeability kc ≤ 1 mD) can prevent methane and brine leakage over a time scale of up to 100 years. However, a poorly cemented borehole (kc ≥ 10 mD) can yield methane leakage rates ranging from 0.08 m³/day to more than 100 m³/day, depending on the hydrodynamic properties of the target formation and the quantity of mobile gas after abandonment. These leakage rates are consistent with Surface Casing Vent Flows (SCVF) reported for 952 wells (among 1019 wells with positive SCVF) examined between 1995 and 2013 in British Columbia. Simulated long-term brine leakage rates after 100 years for poorly cemented boreholes (kc ≥ 10 mD) are on the order of 10-4 m3/day (0.1 l/day). Based on scoping calculations with a well-mixed aquifer model, these rates are unlikely to have a major impact on groundwater quality in a confined aquifer since they would only increase chloride concentrations by about 1 mg/l above background, a negligible increase relative to the commonly recommended aesthetic objective of 250 mg/l for chloride.