Publications

This paper presents the application of probabilistic history matching and prediction workflow in a real field case in Indonesia. The main objective of this novel approach is to capture the subsurface uncertainty for better reservoir understanding to be able to manage its risk and make a better decision for further field development. The field is very complex, with updated geological concept of multi-level reservoirs that has more than a hundred of wells and has been producing for 70 years. Existing multi-realization of static reservoir model was built to determine range of probabilistic cases of In-Place calculation as output. Variation of fluid contacts, lithology/facies distribution, porosity distribution and Net to Gross map are the main differences among these cases. Structural model and reservoir properties from three pre-defined cases were imported to the integrated software modelling tool, excluding water saturation model. The static-dynamic model building process were then recorded under common workflow for integration and automation of rebuilding variation model. For effective probabilistic model initialization,an automatic capillary pressure adjustment was chosen. Subsequently, experimental design and optimization were run to manage probabilistic history matching effectively. Parameter screening and ranking tool were also used to update uncertainty design for the next iteration. The number of history match variants were managed by applying acceptable match criteria and clusterization. Twenty equiprobable history matching variants were selected to be carried over to prediction phase and the three selected remaining oil saturation distribution maps were assessed for waterflood pattern design. Having reduced the uncertainty of parameters by history matching process, the prediction of base case and waterflood scenario were run for twenty unique variants. Incremental cumulative oil is in the range of 14.81 MMSTB to 16.96 MMSTB, equivalent to incremental recovery factor 5% to 5.4%. This range represents static and dynamic input parameter uncertainty that examined in this study. High side of recovery factor from waterflood scenario is 21.6% which indicates many remaining unswept oils. These results were used for work activity recommendation in the future to recover more hydrocarbon from the 70 years old oil field. This paper demonstrates the first application of probabilistic dynamic modelling in the company including a first-step endeavour to integrate static and dynamic variable uncertainty for this field. The workflow will be used as a guideline process for other field applications in the future.