RTA-Assisted Numerical History-Matching and Co-Optimization of CO2 Storage and HnP Performance for a Near Critical Gas Condensate Shale Well

Due to strong non-linearities in the diffusivity equation, numerically-assisted rate-transient analysis (RTA) techniques have been suggested for analysis of multi-phase production data from multi-fractured horizontal wells (MFHWs). However, these methods are based on some limiting assumptions that cannot be generalized for three-phase flow or when relative permeability is unknown. In this study, a new RTA-assisted history-matching technique is proposed to simultaneously match production data and diagnostic plots during the calibration process. In the proposed method, the objective function is modified to include the derivative of the integral of rate-normalized pressure for the primary phases. As such, in the history-matching process using compositional numerical simulation, the flow regimes are also matched, which can increase the reliability of the calibrated numerical model. This approach is applied to a challenging dataset: production data from a MFHW completed in a Canadian shale reservoir hosting a near-critical gas condensate fluid. The calibrated model is then applied to co-optimize CO2 storage and oil production using a cyclic gas injection scheme. The results demonstrate that when the modified objective function is used, the history-matching scheme will reject models that cannot reproduce the flow regimes even if the production data are visually matched. Another benefit of this modified history-matching workflow is that, unlike other numerically-assisted RTA techniques, it is not limited to any specific conceptual model or reservoir geometry. Further, interactions between parameters are accounted for during the calibration process. Co-optimization using the calibrated model leads to an optimized Huff-n-Puff (HnP) design that can produce 40% additional (incremental) oil, while around 17% of the injected CO2 is stored during the cyclic CO2 injection process. In this study, a modified objective function is introduced for the first time to enhance the numerical history-matching process to ensure the resulting calibrated model can also reproduce the observed transient flow regimes. This approach is easy to implement and is not limited to a specific model geometry or any input-output relationship.