An Integrated Workflow for Data Analytics-Assisted Reservoir Management with Incomplete Well Log Data

Summary Advancements in data collection, storage, and processing have facilitated the widespread adoption of data-driven models for forecasting and decision-making in the oil and gas industry. Earlier studies suggested that when we have a limited understanding of reservoir characteristics and physics, data-driven modeling approaches can potentially contribute to better engineering and management of subsurface energy resources. In this paper, a data-analytics-driven, integrated workflow for reservoir management is presented. The workflow is applied to a real oil field with 48 wells, which have been producing from three fractured-carbonate, undersaturated reservoirs. After data compilation, multivariate imputation using chained equations (MICE) methodology is applied to estimate missing well log sections. The complete data set is then used to obtain spatial estimations of reservoir properties throughout the reservoir area. A machine learning–based performance forecasting model is designed, developed, and validated using available data at well locations. This model is used to extend the performance forecast to undrilled locations in the reservoir to identify potentially promising regions for field development. Model results revealed central and southeastern sections of the studied field as high-potential regions for additional oil recovery, which was consistent with earlier geologic interpretation. Although individual parts of the presented workflow were discussed individually in different papers, the main novelty in this study is developing and applying an integrated, field-data-driven workflow that starts with incomplete raw data and ends with field development recommendations. Dealing with multiple scales of static and dynamic types of data for reservoir management remains as a challenging task in complex reservoir systems, and such kind of a workflow can integrate available data for better decision-making and field development practices.