“…Various methods of simulating fracture distributions and networks have been proposed during the last three decades, the orientations of fractures are generally presented: as random or directly related to fracture sets (Long et al, 1985;Andersson et al, 1984;Andersson and Thunvik, 1986;Chilès, 1988;Billaux et al, 1989;Cacas et al, 1990;Jafari and Babadagli, 2012), by assuming probabilistic functions such as those of Fisher (Min et al, 2004;Xu and Dowd, 2010), or by a stochastic realization within a tolerance angle for each dominant fracture set (Dowd et al, 2007;Rafiee and Vinches, 2008). A realistic simulation requires the conditioning of the orientation data for observed fractures, in addition to the position and length, because spatial correlation of the orientation has been shown to exist (i.e., similar orientations tend to appear at nearby fractures) by La Pointe (1980) and Guo et al (2009). Following Koike et al (2012), we apply a method, GEOFRAC (the GEOstatistical FRACture simulation method) that has merit in its extension to 3D modeling and suitable assignments of fracture density and orientations.…”