F. V. A. Risso1, V.F. Risso and D. J. Schiozer
A risk analysis can be applied to several phases of the development of a petroleum field. A typical analysis requires many realizations. Selection of the realizations is usually made in a way that maximizes the coverage of anticipated range of input parameters while requiring the minimum number of simulation runs. Techniques such as Latin Hypercube, Statistical Design, Derivative Tree and Monte Carlo can be used in this process. Alternatives to speed up the process are the reduction of the number of attributes and their discretization levels or the substitution of the conventional reservoir modeling by faster techniques (proxy models). The main objective of this paper is to apply and compare the precision of the results using three different techniques. Derivative Tree, Monte Carlo, Latin Hypercube. The idea is to reduce the number of simulation runs maintaining the precision of the results. The case studied represents a field with 4 main uncertain attributes: horizontal and vertical permeability, porosity and net pay. The objective functions analyzed are net present value, oil recovery factor, water production and oil production. It is shown that the choice of the right technique in this paper allows a reduction in the number of simulations, maintaining the precision of the risk curves. It is shown that the results were very satisfactory with the application of the Latin Hypercube Technique. The results showed that, in this case, this technique is more adequate than Derivative Tree and Monte Carlo considering the required number of simulatins.
Share this article