The use of uncertainty analysis as a tool in reservoir studies is becoming more and more common inside Petrobras and all around the world. However, in fields with production history, traditional uncertainty analysis, combining possible values of uncertainty variables, can lead to models that poorly represent the reservoir and to results that do not respect the available dynamic data. During uncertainty analysis process, history matching evaluation can considerably reduce the existing uncertainties.
The methodology used in this work is based on experimental design and response surfaces. Besides the cumulative production response surface, another one is generated to represent the quality of the history matching. Only cases with a good history matching are selected as input to the Montecarlo simulation. With this technique, it is possible to evaluate the initially defined probability distributions and, if necessary, to redefine shape or limits for the probability density curve.
The methodology was applied in a real study in Petrobras. There are uncertainties related to faults, absolute permeability and also related to the existing fluid properties. Although there are other wells in the same block, the studied area is located in a sea-bottom slope region, where water depth varies considerably within the block, possibly influencing the oil quality.
Since there are two wells operating in the studied region, one producer and one injector, the developed analysis took the existing dynamic data into account, reducing model uncertainties.