The Era of Brutalist Subsurface Engineering, Why the Time is Now

Abstract A challenging task faced by subsurface teams is to decipher reservoir flow behavior based on limited/insufficient information on relevant geological, fluid dynamic and thermodynamic variables. The conventional approach consists of integrating multiscale data into software platforms to create a digital clone of nature. This is followed by attempts to model reservoir reactions in response to human interventions. As the scale narrows from field to well scale, challenges and uncertainties escalate due to the intricacies of near-wellbore scenarios, marked by wellbore features, high fluid velocity and mobility effects. This paper showcases an alternative workflow that aims to create simplified and fully functional models focused on flow processes rather than imaging the reservoir. The workflow is based on the following structure: identify key geo-denominators that drive the reservoir process in question, model the simplified process by testing geo-variables individually or in groups of interdependent variables, rank geo-variables based on their impact on the production signature as determined in the previous step and determine the most influential, flow-critical geo-denominators, create a catalog linking geo-variables to production signatures, run the actual reservoir process through the catalog and nominate the best matched behavior as the closest approximation to the reservoir feature. Step (v) can be assisted with machine learning algorithms. This method systematically evaluates key geo-variables and their contributions to observed flow patterns, leading to more insightful understanding reservoir dynamics. In summary, these are the five benefits of the approach: Cost-Effective and Efficient Modeling: Enables the creation of inexpensive and quick models to capture key reservoir geological architectures and flow-dominating features. These models are effective with limited datasets and do not require extensive history matching. Risk Assessment and Physical Proxies: Facilitates the design of physical proxies and conducts physics-based risk assessments, providing critical information for informed decision-making. Standalone and Integrative Tools: Serves as standalone tools for reservoir studies or can be integrated into workflows for full-field modeling. They are ideal for evaluating "worst-case" scenarios, validating analysts’ intuition, and supporting informed decision-making. Data-Driven Insights: Provides critical information necessary for making informed decisions, ensuring a comprehensive understanding of reservoir behavior and geo-parameter interactions. Integration with Machine Learning: Abstract models naturally integrate with machine learning algorithms, boosting predictive power and enhancing decision-making in reservoir management strategies. The method has been successfully applied in a fractured carbonate field in the Middle East and one application is detailed in this paper. The proposed approach delivers value where the computational cost or the impracticality of detailed, high-resolution models are of concern. By removing non-essential details for flow, engineers can gain quick insight into reservoir behavior enabling them to mature decisions for the optimization of recovery. This approach proves particularly beneficial at well scale and supports WRM applications, where processes develop rapidly, and subtle geo-features can exert a substantial impact on flow behavior.