Abstract
The multi-stage fracture treatments create complex fracture networks with various proppant type, size, and concentration distributed within and along fractures through reservoir rock, where larger size and higher concentrations usually result in higher long-term conductivity. To model the fracture conductivity reduction with depletion, we traditionally use a single monotonic relationship between fracture conductivity and pressure, which is proper for a single proppant concentration but obviously hard to describe the situation in the horizontal wells with complex concentration distributions. This paper is to present a new method to speed-up the calibration process of well performance models with multi-million cells and its two applications in the Wolfcamp reservoir in the Delaware Basin.
To study well performance and completion effectiveness of 3000 horizontal wells over University Lands acreage in the Permian Basin, we have built a series of well performance models with complex fracture networks (SPE 189855 and 194367). We have used those models to methodically investigate the drivers of well completion parameters and well spacing on well performance and field development value (URTeC 554). In the process of building multiple robust well performance models, we found out it is hard and time-consuming to calibrate a well performance model with multi-million cells based upon a single correlation between fracture conductivity and pressure.
We first modeled the complex fracture networks and fracture conductivity distributions based upon the historical completion pumping data; we then developed multiple correlations to characterize fracture conductivity reduction and closure behaviors with pressure depletion based upon initial fracture conductivities (as the result of proppant type, size, and concentration) and reservoir geomechanical properties. We found out that this method significantly reduced our model calibration time. We then applied our method to multiple case studies in the Permian Basin to test and improve the method.
We have thus developed a method to mimic the fracture conductivity reduction and closure behavior in the horizontal wells with complex fracture networks. The paper will layout the theoretical foundation and detail our method to develop the multiple correlations to model fracture conductivity reduction and fracture closure behaviors in the horizontal well performance models in the unconventional reservoirs. We will then show two case studies to illustrate how we have applied our method to speed up the model calibration process.
Based upon the multiple applications into our model calibration process, we have concluded that the method is very effective to calibrate the well performance model with complex fracture networks.
The method can be used for engineers to simplify and speedup calibrating horizontal well performance models. Therefore, engineers can more effectively build more robust well performance models to optimize field development plans in the unconventional reservoirs.