IYANGBE JOSHUA ESOSA

Traditional reservoir forecasting methods provide single deterministic predictions that fail to capture subsurface uncertainty. This research develops a Python-based simulation framework integrating material balance physics with Monte Carlo uncertainty quantification for probabilistic reservoir performance forecasting.

The framework generates P10/P50/P90 forecasts through efficient tank modeling, completing 5,000-realization uncertainty analysis in under one hour—a 50-100x speedup versus commercial simulators. Application to a representative sandstone reservoir yields 18.5 MMstb cumulative recovery (P50) with ±24% uncertainty (P90: 14.2 MMstb, P10: 23.1 MMstb). Sensitivity analysis identifies permeability as the dominant uncertainty driver (correlation +0.68), directly informing data acquisition priorities.
Validation confirms material balance closure within 0.5% and recovery factors consistent with published analogs. By eliminating software licensing costs and computational barriers, the framework democratizes probabilistic forecasting, transforming reservoir engineering from deterministic prediction to accessible, risk-aware decision support.