Probabilistic Assisted History Matching Approach for Production Prediction and Flow Assurance Assessment

Abstract The behavior of fluids in an oil well is strongly influenced by pressure, temperature, fluid composition variations, and other fluid flow properties, which can impact production performance and pose significant flow assurance challenges. These challenges can compromise the integrity of the production system and disrupt operations. Traditional deterministic approaches in multiphase flow simulation often fail to capture uncertainties in key parameters, such as fluid properties and operational conditions. To address this limitation, this study integrates probabilistic methods into a one-dimensional multiphase flow simulator using an Assisted History Matching framework. The proposed workflow begins with a global sensitivity analysis to identify the most influential parameters, followed by a Bayesian history matching to refine probability distributions, improving model calibration with historical field data. Finally, the uncertainty analysis provides credible intervals for key output variables, bringing more information than traditional deterministic methods. This probabilistic approach enhances the analysis because of the associated uncertainty considered in all analysis workflow. A case study involving two Brazilian wells demonstrates the methodology's effectiveness. Both wells were used to validate the global sensitivity analysis, and Well B was used to validate the history matching and uncertainty analysis. The results show that incorporating probabilistic methods into multiphase flow modeling improves prediction accuracy, enables robust flow assurance assessments, and supports better decision-making in production system design and operation. The results for gas, oil, and water flow rates are close to the historical field data used to calibrate the model in the Bayesian history matching.