Multi-Phase Rate Transient Analysis Considering Complex Fracture Networks

Abstract

Abstract Application of horizontal wells and hydraulic fracturing achieves commercial productivity of unconventional oil and gas resources. Complex fracture networks (CFN) provide flow channels and significantly affect well performance in unconventional reservoirs. However, traditional rate transient analysis (RTA) models barely consider the effect of CFN on production performance. Besides, the impact of multi-phase flow on rate transient behaviors is still unclear. Neglecting these crucial effects could cause incorrect rate transient response and erroneous estimation of well and fracture parameters. To fill this gap, this paper investigates the multi-phase rate transient behaviors considering CFN, and tries to investigate in what situations the multi-phase model should be used to obtain more accurate results. Firstly, an embedded discrete fracture model (EDFM) is generated instead of local grid refinement (LGR) method to overcome the time-intensive computation performance. The model is then coupled with reservoir models using non-neighboring connections (NNCs). Secondly, eight cases are designed using the EDFM technology to analyze the effect of natural fractures, formation permeability, and relative permeability on rate transient behaviors. Thirdly, Blasingame plot, log-log plot, and linear flow plot are used to analyze the differences of rate transient response between single-phase and multi-phase flow in reservoirs with CFN. For multi-phase flow, severe deviations can be observed on RTA diagnostic plots compared with single-phase model. The combination of three kinds of RTA type curves can obviously characterize the differences from early to late flow regimes and improve the interpretation accuracy as well as reduce the non-unicity. Finally, field application in Permian Basin demonstrates that multi-phase RTA model and type curves are required for analyzing the production and pressure data since single-phase RTA analysis will lead to big errors for interpretation results.