You would have to have been living in a cave over the last few years to have missed the “dash for gas” for both tight gas and shale gas plays. At the heart of the success of these plays has been the development of hydraulic fracture stimulation techniques, which provide an efficient means to drain reserves that are otherwise locked in micro-Darcy (or less) permeability rocks. When the success of a well depends on the success of its hydraulic fracture, it is essential to have the ability to predict hydraulic fracture geometries and drainage volumes. This is particularly important in key resource plays such as the Barnett, Marcellus, and Haynesville shales in the US and the Horn River Basin in British Columbia, Canada.

A DFN model built around a well showing the part of the population of simulated fractures built up from well data, with the fractures coloured by permeability.aperture product (KH). (Images courtesy of Golder Associates)

Hydraulic fracture propagation is controlled by a combination of in situ stress, reservoir pressures, the rock matrix, and the natural fracture system near the well. There are two keys to understanding and modeling hydraulic fracture growth. The first concerns the challenges in the dynamic numerical simulation of hydro-fracture growth as this represents a highly complex geomechanical system. This remains at the forefront of modeling capability due to the difficulties in the coupling of frac and pore pressure evolution and the rapid change in fracture geometry as fracs propagate.

The second major issue is how to interpret the evolving microseismic signals in terms of both hydraulic fracture growth and their interaction with the natural fracture system. Geologists and geophysicists attempt to interpret the pattern of events to understand possible frac lengths. However, without a map of possible fracture pathways, these interpretations can be overly simplistic and fail to capture the nature and interaction of the fracs.

Those fractures within the DFN model that are indirectly connected to the perfed interval and stimulated during fracing are shown in red.

To address these key difficulties, during the past two years a discrete fracture network (DFN)-based approach has been developed within Golder Associates’ FracMan code to model and predict hydraulic fracture development within a network of existing natural fractures. This rule-based approach bypasses many of

the above difficulties. The DFN approach has been verified by comparison of simulated microseismic responses to field measurements, by validation against detailed geomechanical modeling, and by verification of predicted tributary drainage volumes against production histories.

On the left is actual and right simulated microseismic events associated with three frac stages. It is the general pattern and extent of the microseismic events that is being matched.

DFN modeling is a 3-D modeling technique that provides a virtual geology of natural fractures constrained by geologic structure, stress-strain history, and lithostratigraphy. It is a discrete modeling approach that builds up networks of individual fractures whose geometries (size, orientation, intensity, etc.) are derived from analysis of available well and surface data. The simulated discrete natural fractures control simulated hydraulic fracture propagation according to empirical rules that mimic how natural fractures control hydraulic fractures in nature — by providing leak-off paths for frac fluid, by providing preferred locations for hydraulic fracture propagation, and by developing a reactivated fracture network for delivery of gas to the well bore.

When to use the model

The level of detail and uncertainty in a DFN model is related to where a field is within the overall development life cycle. For exploration prospects, the DFN model is populated based on geologic understanding, geophysical and reconnaissance data, and analog fields. For fields in early stages of development, this can be supplemented by 2-D seismic, wireline geophysics, and fracture image logs. Mature fields can be modeled in detail, integrating the full range of geologic, geophysical, and hydrodynamic data.

Tributary drainage volume calculations can be determined for each well by identifying the connected stimulated fracture population for each frac stage.

Because the DFN hydraulic fracture modeling approach uses a rule-based approach, it can simulate the propagation of hundreds of hydro-fracs per hour while honoring the in situ stress; rock fabric; and frac specifications such as pressure, flow rate, skin condition, and duration. By using fracture cluster and network analysis tools, these simulations can be rapidly searched to determine the interaction between natural and hydraulic fractures and therefore predict the path of frac fluid and proppant. DFN hydraulic fracture modeling determines the state of stress on each fracture to evaluate the potential for shear or tensile failure of each natural fracture that takes frac fluids. The aperture and permeability of these critically stressed fractures can then be adjusted, accounting for both the shearing of natural fractures and the effect of proppant intrusion.

To help calibrate the model, this DFN approach simulates microseismic events by identifying locations within the natural fracture network and the growing hydraulic fractures where the effective stress exceeds the strength, as defined by an appropriate strength criterion. Comparison of patterns of simulated microseismic events to field measurements increases confidence in the DFN approach and allows the tuning of key hydraulic properties as part of the model calibration process. In addition, the DFN hydraulic fracture approach can be confirmed by comparing the results of the stochastic simulation against detailed geomechanical models such as the hybrid finite element/distinct element code, ELFEN.

DFN studies of a number of shale

gas reservoirs have demonstrated three key interactions between natural and hydraulic fractures: Firstly, intersecting natural fractures bleed frac fluid and proppant, substantially reducing the extent of hydraulic fractures. Secondly, natural fractures intersecting the well bore provide a preferential location for hydro-fracs such that hydro-fracs can initially extend from natural fracture locations, extending and improving the connectivity of those fractures. Finally, reactivated natural fractures (i.e, those that take proppant) can extend the drainage depth of the hydraulic fracture beyond the actual hydro-frac. Repeated simulation operation for each frac stage provides a calibrated DFN that can reproduce the observed microseismic data, the tributary drainage volume, and estimated ultimate recovery of the well. This information can then be used to optimize frac designs, well spacing, and well trajectories. The analyses are numerically efficient, so that Monte Carlo

simulation can be used to quantify the uncertainty (p10, p50, p90) of ultimate recovery for alternative development strategies. Proper consideration of natural fracturing is key to successful development of shale and tight gas resources. The DFN-based approach captures the geometry, hydrodynamics, and geomechanics of natural fractures to increase the success of these resources.