With an increasing global energy demand, optimal enhanced oil recovery strategies are of paramount importance. However, oil recovery strategies depend on unknown rock properties of a subsurface oil reservoir. Production wells provide measurements of the rock properties but indirectly and only at a few locations. Moreover, the rock properties cannot be uniquely defined due to the nature of the measurements.
Therefore, inferring an ensemble of the rock properties given a few indirect measurements from production data is essential to develop optimal enhanced oil recovery strategies. Data assimilation is a widely-used mathematical methodology in meteorology to combine measurements and computational weather models to give a weather forecast. However, in meteorology uncertain fields evolve over time while the rock properties in geoscience are stationary.
This thesis adopts existing data assimilation methods from meteorology and introduces a novel data assimilation method to estimate the rock properties of a subsurface reservoir. The adopted methods give a good estimation when uncertainties are small. However, they struggle when facing large uncertainties in the rock properties. The novel data assimilation method outperforms the standard data assimilation approaches for complex subsurface structures that include channels with different types of rocks and large uncertainties. Though the novel data assimilation method requires a computationally affordable ensemble of reservoir models, by itself it remains computationally expensive. The future development focuses on investigation of computationally cheaper extensions of the introduced method and its application on more realistic scenarios.