A New Numerical Well-Test Model Using an Analytically Modified Embedded Discrete Fracture Model

In the past few decades, multi-stage hydraulic fracturing technology has emerged as a crucial tool for the commercial development of unconventional oil and gas resources. Accurately characterizing transient flow near fractures is a critical concern for many researchers. Currently, discrete fracture models (DFMs) are primarily used to analyze the pressure transient behaviors of multi-stage fractured horizontal wells (MFHWs). Although discrete fracture models can accurately capture transient flow around fractures, they require a substantial number of grids to ensure computational precision, which in turn leads to higher computational costs. Conversely, standard embedded discrete fracture models (EDFMs) based on pseudo-steady-state assumptions, while computationally efficient, struggle to precisely depict the early transient flow around fractures. To narrow this gap, we proposed a new numerical well-test model for analyzing the pressure transient behaviors of MFHWs using structured Cartesian grids and an analytically modified EDFM (AEDFM). We have made modifications to the transmissibility between the matrix and fractures by multiplying it with a transient factor. Furthermore, we have validated the accuracy and efficiency of our proposed model through comparisons with results from analytical models and standard well-test software. This demonstrates the significance of our proposed model in accurately capturing transient flow around fractures and reducing computational costs. Additionally, we conducted research on the pressure transient behaviors of a MFHW under different parameters and further evaluated the significance of the proposed modifications based on the results. The results indicate that, compared to the standard EDFM, the AEDFM can effectively match the early nonlinear pressure drop near fractures. This study may potentially provide a powerful tool for the precise analysis of pressure transient behaviors in MFHWs, while also significantly reducing computational costs.