Uncertainty-Centric Ensemble Modelling and History Matching of Field S: Enabling Robust Redevelopment Planning and Economic Evaluation

Abstract Field S, located approximately 150 km offshore Terengganu, Malaysia, is an under-developed field that has been producing for 13 years. Currently developed with only 15 wells for 12x6 km area, the field is poised for a significant redevelopment expansion, with plans to drill up to 30 additional infill wells over the next five years. Given the substantial capital investment associated with this development plan, a robust understanding of subsurface uncertainties was critical to ensure its economic viability and to manage reservoir risks effectively. With a tight timeline towards project sanctioning and to address this challenge, an ensemble-based reservoir modelling approach was adopted. The primary objective of the study was to implement an uncertainty-driven workflow that could integrate multidisciplinary data, generate probabilistic forecasts, and provide insights into reserves distribution and development robustness that ultimately support well informed decision-making. A comprehensive ensemble of 100 reservoir models was constructed to capture static and dynamic uncertainties. These models incorporated variations in well log interpretations, depositional facies, rock types, petrophysical property distributions, fluid contacts, capillary pressure thresholds, SCAL datasets, and PVT characteristics. The initial ensemble was conditioned to historical production data using an ensemble Kalman smoother based data assimilation technique. Several iterations of the history-matching process were performed, applying a weighted objective function to prioritize the quality of match against observed production metrics such as oil rate, water cut, and bottom hole pressure. Upon achieving acceptable convergence, the cases in the ensemble were ranked based on history match quality, and representative models were selected for forecasting future production under the proposed infill drilling scenarios. These forecast simulations were run across the ensemble, resulting in probabilistic production profiles. The ensemble Kalman smoother-based data assimilation method yielded a reasonable history match across the entire ensemble. Notably, uncertainties were significantly reduced in the regions around producing wells, while areas with limited or no production data retained a broader posterior uncertainty range. From the cumulative distribution function, three representative models capturing the P10, P50, and P90 outcomes of cumulative oil production were selected as technical input for project economic evaluation. The ensemble forecasts allowed project stakeholders to quantify development risk from reservoir volumetric standpoint and make more resilient investment decisions. This study demonstrates the practical value of ensemble-based modelling in a real-field context, highlighting its role in managing uncertainty during field development planning. Beyond achieving a successful history match, the workflow established a repeatable and scalable methodology that supports rapid model updates as new data becomes available. By maximizing the value of existing subsurface and production data, the approach minimizes the risk of overfitting and enables robust decision-making under uncertainty. The insights generated are instrumental in supporting high-stakes economic decisions for Field S future development.