Producing from shale formations has been made profitable because of technological advancements of hydraulic fracturing and horizontal drilling. However, the complexity and uncertainties of the shale reservoirs make it hard to estimate the assets and maximize the value by optimizing the multi-well placement. Reservoir simulation is a powerful tool to estimate the performance of reservoir but calibrating models and optimizing the well spacing can take lots of human efforts and computation time, especially when we need multiple models to stress the uncertainties. Thus, an efficient way to improve the efficiency of simulations and reduce the number of simulations needed can be helpful for the decision-making. In this study, a probabilistic well spacing optimization workflow was developed that can help engineers understand uncertainties and ease decision making process. The EDFM (embedded discrete fracture model) method was applied to efficiently model hydraulic and natural fractures in shale reservoirs. The different combinations of uncertain reservoir and fracture parameters were effectively captured through performing assisted history matching. We present an example of well spacing optimization in a shale gas reservoir with complex natural fractures.
The optimal well spacing is crucial for the economic development of unconventional reservoirs. Disproper well spacing will reduce oil and gas production, as well as economic benefits (Yu and Sepehrnoori, 2018). Therefore, it is essential to find an optimal well spacing that can balance Estimated Ultimate Recovery (EUR) and economic evaluation for hydraulically fractured horizontal wells in shale reservoirs. Several studies have focused on well spacing optimization numerically and analytically. However, because of the shale reservoirs' complexity and uncertainties, it is challenging to accurately calibrate the subsurface uncertainties associated with hydraulic and natural fractures (Chang et al., 2020). Assisted history matching is a good method to capture the realization of uncertainties (Li et al., 2020).
Another challenge is how to appropriately model the complex fractures network, which plays a critical role in gas recovery (Yu et al., 2018). The actual fracture geometry is complicated, especially when natural fractures exist (Sepehrnoori et al., 2020). Embedded Discrete Fracture Model (EDFM) was proposed to overcome this issue. It can save more than 90% time than the Local Grid Refinement (LGR) method while maintaining accuracy (Xu et al., 2017a, 2017b, 2018). Therefore, EDFM is the best approach to accurately and efficiently establish natural fractures and hydraulic fractures in gas reservoirs with high flexibility.