Efficient Modeling of Unconventional Well Performance with Millions of Natural and Hydraulic Fractures Using Embedded Discrete Fracture Model EDFM

Abstract

The complexity of dynamic modeling for naturally fractured reservoirs has increased in recent years to incorporate more data and physics, as well as to handle advanced completion designs and development scenarios. While these complex models can provide more insight to difficult problems, they come with higher computational costs. Such a limitation prohibits an asset team from working with a large number of well/fracture scenarios that correctly represent geological uncertainty. This study presents a powerful non-intrusive Embedded Discrete Fracture Model (EDFM) method to efficiently handle millions of natural and hydraulic fractures with hundreds of horizontal wells, which has never been modeled in the literature. Specifically, we built a 3D geological model using a black oil reservoir simulator with 100 square miles in the horizontal area and 11 layers of 165 ft thickness. The total number of matrix cells without considering fractures is over 3 million. In total, 400 horizontal wells with well length of 6000 ft were modeled in two target layers. Each layer contains 200 wells. Each well has 112 hydraulic fractures with cluster spacing of 50 ft. The total number of hydraulic fractures is 44,800. In addition, we generated three cases with 10K, 100K and 1 million 3D natural fractures with dip angle from 70 to 90 degrees. For the case with 1 million natural fractures, the total number of cells is over 42 million. Well performance for the field example, with and without natural fractures, was investigated. This work adds significant value to the well and fracture spacing optimization process during field development planning. The non-intrusive EDFM method has been proven to be an efficient fracture modeling tool for simulating million-level complex hydraulic/natural fractures, which significantly improves accuracy and reduces computational time.