Jornal de Geologia e Geofísica
Acesso livre

Abstrato

A Comparative Study of Geostatistical and Statistical Methods to Simulate Synthetic Fracture Networks

Sima Shakiba, Faramarz Doulati Ardejani

Estimating the accurate fracture network based on observation is of paramount importance, especially in the field of hydrogeology and mass transport. An incomplete spatial distribution of fractures is the primary cause of uncertainty related to fractured network simulations. In this regard, employing suitable approaches can lead to an accurate simulation of a fracture network. In this paper, a comparative study of two different methods is implemented to deal with the uncertainties related to reproducing fracture length, direction, and pattern. The direct sampling (DS) algorithm, as the newest multiple-point geostatistical simulation, based on sampling from a training dataset is used. In addition, statistical simulation of a fracture network, as the second method is employed to simulate the density, the orientation as well as the length distribution of fractures. The results show that the DS algorithm greatly estimates the fracture orientation and patterns rather than the length distribution. Therefore, in DS algorithm, the total number of small fractures will be increased, and the length distribution with a normal distribution in reference models of fracture zones will change to a log-normal distribution in DS algorithm. On the other hand, the statistical method is capable of reproducing the length distribution of the fracture network. Despite the DS algorithm, the accuracy of statistical simulation in reproducing fractures pattern is not satisfying.