Uncertainty analysis by model selection technique and its application in economic valuation of a large field

Uncertainty assessment for economic valuation of a field generally involves developing multiple subsurface realizations combined with different development scenarios, thereby rendering the task computationally expensive and time consuming. In addition, the task of updating the models with new information is equally daunting leading to enormous and unreasonable simplifications to the economic valuation procedure. This paper presents a novel approach that renders the uncertainty assessment less cumbersome and more efficient/accurate with the help of a new uncertainty assessment approach. The proposed approach makes use of a model selection algorithm to reduce the number of subsurface realization necessary for uncertainty assessment. A model selection framework does the work of refining an initial suite of reservoir models to a final set of reservoir models that depict production characteristics close to the observed field history. Model selection algorithm attempts to group reservoir models based on the common connectivity characteristics exhibited by them and then retrieve the group exhibiting characteristics that are closest to the observed response for a reservoir. The model selection technique is thus an efficient way to reduce number of realizations and save computational time for doing uncertainty analysis, and yet be optimally constrained to the available data. An illustrative example with a dataset for a large field is used to show the proposed methodology for economic valuation of a large field. Well testing data from two wells is used as two sets of history data for the model selection procedure. Real option valuation (ROV), a probabilistic approach, is used to infer the economics of the reservoir. The case described will demonstrate that the proposed approach is systematic, reduces the time of computation, and allows the rigorous application of economic analysis without making any undue simplifications and assumptions.