Uncertainties analysis methods for hydrocarbon volumes estimation

摘要

Uncertainties in the estimates of hydrocarbon volumes are traditionally given by the Monte Carlo method, which is based on classical probability theory. There are a lot of uncertainties, for example: the geological model, physics and thermal parameters, lithologies, dates, etc... An alternative way to achieve the same objective is to use arithmetic. This approach is based on possibility theory derived from fizzy set theory. The fuzzy mathematics is attractive because, unlike the Monte Carlo Method, it does not have to assume the statistical distributions of geological variables, and also because of its simple computations. Furthermore, in the cases that the distributions are known, there are a correspondence between the two formulations. When the Monte Carlo method must be adopted, a neural network can be trained to reduce the number of samples required, and consequently the total computer time consumed, that can be prohibitive in 2D and 3D simulations. In this paper these aspects are focused and some comparisons are made to explicit the main advantages that are obtained in practical situations.