Integrated Uncertainty Quantification for Development Planning of a Large Field

Abstract

Oil companies invest significant human and money resources on decisions about field development planning. Often these decisions have to be made based on limited information under tight time constraints. Historically, numerical simulation is one of the main tools in the process. In most cases, the investigations remain limited to single realizations or at best for several representative model cases. However, in order to make reliable decisions about field development the whole range of possible realizations has to be taken into account.

Ideally, a complete study of production forecasting has to include all types of uncertainty in the reservoir: the structure model, reservoir property distributions, fluid properties, well parameters, etc. But unfortunately an extensive volume and complexity of simulations, limited project time, limited hard and production data, great number of uncertainties, and wide range of initial fluid properties make the problem extremely challenging. In reality even detailed uncertainty studies include only several geological model realizations and a few of the most sensitive parameters at the simulation model level.

In this work, we consider a practical case of an integrated uncertainty assessment study for a large multi-layered oil field with more than 200 projected wells. The workflow applied includes various uncertainties from different conceptual steps: reservoir structure, reservoir properties, absolute and relative permeability parameters in the simulation model and well productivity parameters that represent technological uncertainties. The key objective of the project was to locate conceptually different models that match historical production data and carry out a detailed investigation of the prediction scenarios for the next 25 years on the field level and optimize the development plan based on multiple history matching solutions. The project team has built a unique assisted history matching workflow that allows to have multiple solutions to the problem. As a result of several series of simulations, 83 different realizations of the model with equally good match quality were found. These models were used as the basis for probabilistic forecasting and optimization of various development scenarios with account for uncertainty.