Automated pressure transient analysis (PTA) with real-time data feed from permanent downhole gauge (PDHG) enables continuous monitoring of well and reservoir that facilitates timely surveillance decisions. However, while robust automation of the process is critical to minimize the requirement of manual efforts, a challenge lies in automatic diagnostics of a log-log plot which is often contaminated by non-reservoir response such as wellbore dynamics. We propose a new automatic PTA method to enhance accuracy of diagnostics.
The method utilizes a pattern detection method based on similarity search and automatically identifies sequence of flow regimes, such as radial, spherical or linear flow etc., on a log-log diagnostic plot of pressure and derivative. To discover individual flow regimes, the algorithm scans a window on the plot and finds a pattern that is most similar to a ‘motif’ defined for the flow regime. Such motifs are known for individual flow regimes from analytical models. During the similarity search, the algorithm ensures that the discovered sequence of flow regimes is consistent with the flow scenario anticipated at the well.
The proposed method is implemented in fully automated PTA workflow. First, the system reads PDHG pressure and flow rate at a well. Then, pressure buildup intervals are automatically identified. Subsequently, a log-log diagnostic plot is automatically generated for each buildup and the proposed method is executed. Once a sequence of flow regimes is identified, the algorithm locates a horizontal line over the radial flow regime and calculates permeability, skin and extrapolated pressure p*. For horizontal wells, effective completion length is also computed by locating a half slope line on the linear flow regime. For hydraulically fractured wells, fracture length or fracture conductivity is estimated from the linear or bi-linear flow regime. The results are written on output files or to a database together with identified flow regimes visualized on plots for the review of reservoir engineers. The method is tested on oil producers with high water cut where significant fluid segregation or crossflow is impacting log-log diagnostic plots, as well as gas wells where a pressure leak during buildup is contaminating pressure derivatives. Despite such noise of non-reservoir responses, the proposed method successfully identifies flow regimes on most of buildups and produces PTA results comparable to manual analysis.
Compared to existing automatic PTA methods, such as automatic matching of model response or automatic semi-log analysis of radial flow regimes identified by user-specified criteria, the proposed method is particularly robust to use with pressure data which is significantly contaminated by non-reservoir responses. Such robustness of our method is achieved by a flexible pattern search for individual flow regimes rather than matching an entire model response all together or requiring rules specified by engineers.