Fuzzy Logic Assesses Interwell Connectivity: Case Study of a Giant Presalt Reservoir

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 231704, “Using Fuzzy Logic To Assess Interwell Connectivity: A Case Study of a Giant Presalt Reservoir,” by Marcelo A. Gomes, Alisson Alves Da Silva, SPE, and Paulo R. da Motta Pires, SPE, Petrobras, et al. The paper has not been peer-reviewed. _ Carbonate reservoir characterization is challenged by strong heterogeneities and diagenetic features such as fractures and karsts, which affect well-connectivity knowledge critical for drainage and recovery planning. The complete paper applies fuzzy logic, a well-established artificial-intelligence technique, to assess connectivity between injectors and producers quantitatively in a giant presalt field in the Santos Basin of Brazil. Using field data, the fuzzy system fuses qualitative expert insights into a quantitative connectivity index. The results support local model adjustments and fieldwide injection strategies, improving reservoir management. The approach effectively mimics specialist evaluations and enhances tracer data interpretation and geological distribution modeling. Methodology Fuzzy System for Well-Connectivity Assessment. The proposed method is intrinsically data-driven. Instead of predictions from conventional reservoir modeling, the technique relies strongly on available field data, namely production and injected rates and tracer-monitoring surveys. These data manifest features that have a prevalent impact on dynamic reservoir behavior. In processing these data, the goal is to infer the degree of connectivity between injectors and producers. Field data also can be processed and interpreted by human experts, thus providing qualitative insights into the connectivity between wells. In general, this task is performed through visual inspection of charts, maps, well logs, and special-core-analysis tables. The fuzzy system does something similar, but delves deeper into the numerical and statistical structure of the data set. After the data is properly codified and preprocessed, the system consumes them, performs their aggregation, and provides a classification, expressed in numbers or degrees of pertinence to predefined categories, that qualifies the connectivity between pairs of injector and producer wells. When combined with geological interpretations and characterization, the fuzzy system serves both as a basis for local adjustments in the reservoir model and as an insightful source of information for distributing injection rates throughout the field. The method was devised and applied to the central portion of a giant, ultradeepwater, presalt reservoir in the Santos Basin. Five injection wells and 14 producers were chosen for analysis. An assessment of the connectivity between each injector/producer pair was conducted. To cope with massive average reservoir thickness and strong vertical heterogeneities, wells were segmented into zones with intelligent completion installations. The production from or injection into each zone in a given well is controlled by remotely actuated valves. Each zone also is equipped with dedicated instrumentation that allows for individual performance assessment either during regular well operation or as a support for occasional zonal testing. These resources allow reservoir-management teams to better distribute injection rates among the different reservoir layers. Additionally, they help control the gas/oil ratio in producer wells by making it possible to shut off the most-offending zones in each well.