In situ stress field in the Athabasca oil sands deposits: Field measurement, stress-field modeling, and engineering implications

Abstract

In situ stress is one of the key engineering parameters that ensures caprock integrity in steam-assisted gravity drainage (SAGD) operations. Massive steam injections during an SAGD operation leads to significant changes in the pore pressure, temperature, stress, and volumetric strain in the rock formations. These changes trigger deformation and stress redistribution in the overburdened strata, which could lead to a containment breach of the caprock through shear or tensile failure. To ensure subsurface steam containment, both to satisfy public safety considerations and instill confidence that the steam containment will continue over the necessary timescales, the Alberta government requires every SAGD operator to measure the in situ stress conditions in the oil sands reservoir and its caprock. Minifrac test is a suggested method that can measure the minimum principal stress with high confidence. A safety factor (SF) of 80% is used to determine the maximum operation pressure (MOP) for safe SAGD operations; in other words, MOP = 80%*σmin, where σmin is the minimum principal stress at the base of the caprock. In most cases, this 80% SF safeguards the SAGD injection pressure to always remain lower than the σmin at the base of the caprock, ensuring that no tensile fractures are initiated at the caprock.In this study, we compiled a database containing in excess of 100 minifrac datasets across the Athabasca oil sands region and constructed an in situ stress map. Based on our in situ stress dataset, we developed a quantitative risk model to analyze the effects of uncertainties on the probability of exceedance (i.e., the probability that the MOP exceeds the fracture pressure at the base of the caprock). The risk model involves defining probability distributions to quantify the uncertainties at the shallowest depth and minimum stress gradient. The output of the risk model was a set of predictions of exceedance probabilities, which were computed using Monte Carlo simulations. Through a statistical analysis of this in situ database and risk-based simulation analysis, this study demonstrates the reliability of a SF of 80%.