The use of proxy models in risk analysis of oil fields

Abstract

The use of flow simulator in risk analysis of oil fields can require excessive computational time in several steps of productive life of a petroleum field due to the complexity this tool requires, mostly in probabilistic studies, where the number of simulation models can be high. This project aims to develop and apply low accuracy models, also called proxy models, in steps of risk analysis of oil fields, which demand a great computational effort. The use of proxy models can enable the evaluation of a higher number of scenarios, which can increase the chances of better decisions or reduce the computational effort. The methodology proposed in this work involves: statistical methods, statistical planning, response surface methodology and consistence test. It is expected that the implementation of the proxy model suggested in part of risk analysis process submit close results of the numerical simulator of oil reservoir, showing that the use of a mythology integrated to statistical methods, when compared to the flow simulator response, it's possible to accelerate steps of studies containing numerical reservoir simulation and risk analysis and thus, obtain results that allow a quick decision-making without presenting a large margin of error.