Stochastic modelling of wellbore leakage in British Columbia

Wellbore leakage is complicated to understand due to the range of potential leakage pathways and uncertainties regarding their capacities. In this study we present a novel approach to modelling realistic leakage along microannulus pathways of varying thickness. We use stochastic methods to calibrate leakage pathway dimensions to the surface casing vent flow (SCVF) leakage rates reported in British Columbia, Canada. Results shows that representing dry microannulus thicknesses with a lognormal distribution provides a good fit for the intermediate ranges of SVCF flow rates, but that a dry microannulus alone cannot account for all instances of wellbore leakage. We then approach small and high flow rates independently, offering explanations for these. This includes a wet microannulus/mud channel model to account for instances of poor mud removal, which is better able to account for the less frequent higher leakage rates. We conclude that flow rates above 10 m${}^3$/day are progressively likely to be caused by significant failures in mud displacement during primary cementing, or other extreme events such as casing failure due to geological or operational factors.