Fractured reservoirs: a comparison of different modelling approaches, from a well to reservoir and reservoir to well perspective

Abstract

Predicting the potential for successful hydraulic stimulation of shale targets in the Netherlands is key when assessing the optimum areas for well emplacement. The difficulty of assessing areas which will respond favourably to stimulation is associated with limited data sets, namely the absence of well tests and adequate case studies. Therefore, studies have attempted to build robust reservoir models which estimate the fracturing potential of the shale. We compare two modelling approaches. Firstly, an elementary modelling technique where only tensile fracturing is assumed. Secondly, we have developed a semi-analytical model which creates a synthetic natural fault network based on a power law distribution, and established a new method of spatially distributing fractures. This study analyses the petrophysical rock properties at a number of well locations penetrating the Posidonia and Geverik shales. Hydraulic fracturing simulations indicate that the uncertainty associated with leak off has a profound influence on fracture dimensions. We show that areas with a high Young's modulus are characterised by longer fractures and conclude that stimulation is likely to be most successful at well KWK-01 (near Boxtel). Results from the analytical model indicate that reactivated faults following a power law distribution also follow a Guttenberg-Richter distribution. Additionally, moment magnitudes predicted by the model are similar to those recorded during hydraulic fracturing in other studies. We use seismicity within the reservoir to calculate permeability and fluid flow. From this we recreate the reactivation of faults within the reservoir by injecting fluids until yield pressures on outermost faults are reached and record the total volume of fluid. This total fluid volume is empirically related to the cumulative seismic moment with the method of McGarr (2014). We observe a good match between model predicted seismicity and McGarr (2014) predictions showing the validity of the model. The results from the semi-analytical natural fracture model indicate that a power law distribution of faults is a meaningful way in which to represent natural fractures within reservoirs.