**5. The importance of 3D natural fracture models**

Natural fractures have a significant impact in unconventional reservoirs, yet they are rarely accounted for in most physical modeling related to hydraulic fracturing. Natural fractures could have a positive impact as they create additional surface contact during hydraulic fracturing which is commonly referred as fracture complexity [10]. This can be predicted with geomechanical modeling and validated with microseismic response [11, 12]. The contribution of the natural fractures could also be negative by creating direct links to water bearing faults [13] or by creating frac hits [14, 15], through poroelastic effects, that will often damage the production from child and parent wells. Given their importance, a predictive model that provides the 3D distribution of these natural fractures is a critical input for any model trying to predict the outcome of hydraulic fracturing. However, finding a 3D distribution of the natural fractures has two major challenges: how to define the natural fractures at the wells given the rare occurrence of core or image logs in unconventional wells, and how to distribute the limited well data in the 3D reservoir to create a predictive model.

One of the motivations behind the use of surface drilling data is to be able to extract a fracture indicator that can be used to enrich the poor and limited statistics of natural fractures indicators found at wells and using these to build a 3D natural fracture model. Since the natural fractures are not a result of a depositional phenomenon only, their prediction is very different from mapping a more conventional property such as reservoir porosity. Having few limited wells with core or image logs will likely not provide the full statistics of these natural fractures. In other words, most statistical methods such as Discrete Fracture Networks (DFN) may not have the proper statistics from the wells to make any reasonable prediction of their 3D distribution. Attempts were made to reduce this problem by constraining the DFN with a continuous property [16] to guide the statistical distribution but other issues made the use of DFN in natural fracture modeling very challenging. Among these challenges include the dramatic variations found in the upscaled properties [17] needed for additional use of the DFN in engineering applications.

To avoid all the issues related to the use of a DFN, the Continuous Fracture Modeling (CFM) approach was developed to create validated predictive models of natural fractures [8, 9]. The CFM approach takes full advantage of the surface drilling derived fracture index or even the limited statistics that can be found in any natural fracture proxy, image log or core. The CFM approach honors structural geology concepts and focuses on the drivers that influence the presence of natural fractures. For example, the density of natural fractures at a given point in the reservoir does not depend on poorly sampled statistics of various fracture sets measured through limited wireline data, but on the volumetric distribution and interaction of lithology, structural settings and distance to faults, porosity, and many other reservoir properties that create the resulting natural fractures. These reservoir properties commonly called natural fracture drivers could all be estimated directly or indirectly through geologic modeling and seismic processes that involve seismic inversion, spectral decomposition and volumetric curvatures [9] when the data is

*Surface Drilling Data for Constrained Hydraulic Fracturing and Fast Reservoir Simulation… DOI: http://dx.doi.org/10.5772/intechopen.84759* 

available. Since the relationship between the natural fracture drivers and the limited natural fractures available at the wells is complex, artificial intelligence [8] is used not only to retrieve any existing and potential correlations that honors the limited statistics but also all the structural geology concepts. With this approach, extrapolation beyond the limited statistics is possible and has been successfully applied during the last three decades to various problems requiring an accurate description on where the natural fractures are. Among these, are problems in geomechanics such as interactions between hydraulic and natural fractures. This interaction could be better understood if studied in a decoupled way; where the natural fractures role in altering the regional tectonic stress and their impact on the lateral propagation of the hydraulic fracture is separated from the effects of natural fractures during the vertical propagation of the hydraulic fractures.
