Abstract
The economics of tight gas fields highly depend on the consistency between expected production and the actual well performance. A mismatch between the reservoir quality and the well production often leads to a review of the individual well. However, such mismatch may vary from case to case, and it is hard to perform a field-level analysis based on individual well reviews. We introduce a new method based on data mining to assist the field-level diagnosis.
LX gas field is located the in eastern Ordos basin. Compared to the main gas field in the center of the basin, LX field is less predictable in well performance. This predictability issue hinders field development in LX field because the field economics are substantially jeopardized by the inconsistency between the reservoir quality and the production performance.
The traditional workflow to understand this issue at the field level is to review the details of a large number of individual wells in the area. This is typically an intense task, and too much detail from multiple disciplines may hide the true pattern of the field behavior. To resolve this issue, we applied data mining in our field development diagnosis workflow.
Our new workflow in LX area started with the existing field datasheet, including logging summaries, completion treatment reports, and flowback testing datasheets. With the data extracted from these different sources, we visualized the consolidated information in various plots and graphs based on regression analysis, which revealed the relation between flowback ratio and the production, the flowback rate consistency from the different service suppliers, and the impact of water productions.
The data mining approach helped to generate new understandings in LX gas field. With the in-depth analysis of the flowback data together with reservoir properties and operation parameters, the key problems in the field were identified for further development optimization, and the field economics can be significantly improved. The diagnosis method can be easily adapted and applied to any field with similar problems, and data mining can be useful for almost all large-scale field development optimizations.
