Advances in horizontal drilling and new practices in hydraulic fracturing have changed the paradigm of shale reservoirs in the last decade. Nevertheless, completion and stimulation engineers still face serious challenges due to the complex physics involved during hydraulic fracture propagation including hydraulic fracture interaction with natural fractures, stress shadow effects, and proppant transport in complex fracture networks. One of the main questions is how to optimize the number of stages and the placement of perforation clusters accounting for these complex physical phenomena and the wells' economics. To answer this question, it is necessary to analyze how the completion design and the fracturing process are related to the short and long term production.

This paper investigates the relation between the production and the completion design. A state-of the-art, fracturing-to-production simulation workflow is used to carry out a parametric study on completion design. The fracturing simulations are performed with the unconventional fracture model (UFM) that models the hydraulic fracturing process in a complex formations with pre-existing natural fractures including interaction with natural fractures and between hydraulic fracture branches (stress shadow effects). The resulting complex fracture networks are then explicitly gridded to build an unstructured grid that is then passed to a numerical reservoir simulator to run the production simulations and accurately model multiphase reservoir flow around complex hydraulic fracture networks. The base case of this study represents a synthetic reservoir model replicating properties of the Marcellus shale. One of the main parameters investigated is the number of perforation clusters per stage for both a constant pumping rate and for a constant average rate per perforation cluster. We also investigated the influence of the number of stages on production, for a given lateral length and a given total treatment volume. The results from this study provide new understanding of the impact of completion design on production and illustrate its use to find optimum completion design based on modeling. For example, some results show that for a constant average rate per cluster a clear optimum can be found as function of the number of cluster per stage, while this task can be more challenging with a constant total pumping rate.