Depending on the geoscience data available (top blue box), many techniques (middle green box) are available to detect natural fracture systems in the shale reservoir. Fracture detection output (bottom orange box) improves confidence in the reservoir modeling results, ultimately leading to optimal horizontal well planning, fracture simulation, and flow simulation. (Images courtesy of Paradigm)

Unconventional US gas shale play

Though shales contain hydrocarbons, in the past they were always considered as source rock and seals and not as reservoirs. In the United States, an estimated 500-800 Tcf of gas in place lie within gas shale plays. In fact, gas shales may be the last really big onshore play in the United States and could provide 10% of the nation’s gas by 2015 if we can identify and recover these assets.

The Barnett Shale play in northeast Texas has led the learning curve on how to economically produce gas shales. Individual well production output is modest, but collectively, the thousands of Barnett Shale wells have made this the second largest gas field in the United States in just a few years. Not only are Barnett shales thick (100 to 1,000 ft or 30 to 305 m) and organically rich, but they are also brittle and fracture in areas of stress.

Two technologies have made gas shales economical in the past 10 years. One is better fracturing technology. The other is improved horizontal well technology, which increases the possibility of intersecting more fractures.

Figure 1 demonstrates how different geoscience data can be analyzed and interpreted to detect natural fracture systems (NFS). Different techniques are useful to define stress fields, generate fracture density histograms, and provide actual fracture picks from 3D seismic and wellbore information. The results from these NFS techniques can be correlated to improve confidence in the fractured reservoir model.

Multi-azimuth seismic NFS detection techniques

If 3-D seismic has been gathered with azimuthally varying shots and receivers, fractures perpendicular to the azimuth can be imaged better, leading to a more accurate estimate of the stress field and direct fracture detection. The seismic processing workflow involves sectoring the data into azimuthal volumes and independently picking velocities for each cube. One multi-azimuth NFS detection technique involves comparing velocities of different sectors.

AVO isn’t just good for hydrocarbon detection; it can also measure anisotropy. Multi-azimuth AVO (AVOZ) analysis takes advantage of all traces in the gather to provide a more accurate estimate and more confidence in the output stress fields than stacked amplitude analysis. While anisotropy also can be influenced by factors such as fluid and lithology changes, these parameters are constant for thick gas shale reservoirs, and therefore indicative of NFS.

Another multi-azimuth technique compares stacked amplitudes at the same CDP location from different sectored azimuths. After applying statics, an ellipse is calculated to fit the azimuthal amplitude vectors. Stress magnitude and orientation can be established by measuring the eccentricity of the ellipse.

Coherence cube technology can be applied to azimuthally sectored volumes to delineate fractures that are more accurately imaged with perpendicular shot-receiver azimuths. Coherence cube processing is applied to each sectored volume and then automatic fault extraction (AFE) is used to efficiently extract fracture picks and generate rose diagrams.

S-Wave seismic NFS detection

Multi-component surveys make use of downgoing P-waves that convert on reflection to upcoming S-waves, which are effective for natural fracture detection. Quite often, shear waves are more susceptible to fracture or differential stress than P-waves. Open fractures have a big effect on shear wave velocities and amplitudes. The NFS detection workflow first establishes P-wave velocities and Vs-Vp ratios. P and S interval velocities can be determined after pre-stack time migration (PSTM). Not only can the converted wave 3-D volume be directly interpreted for fractures, but Vp/Vs and P-S amplitude ratio analyses can yield direct fault picks as well.

NFS detection techniques for all 3-D seismic

Post-stack enhancement techniques can be used on any 3-D seismic volume regardless of its acquisition.

Coherence is a local 3-D waveform measurement that takes the central one, examines waveforms around it, and derives a cube of similarity to that trace. Variations in neighboring waveforms define low coherence, which is related to faults or stratigraphic discontinuities. Figure 2 shows these low-coherence features in black, with high coherence (unfaulted) features in white. Manually interpreting fractures exposed by coherence cube is tedious and time consuming. AFE quickly generates fracture lineaments and connects appropriate lineaments. Two meaningful outputs are generated by this process: direct fracture picks and rose diagrams of the lineaments that produced the fracture picks. The bottom image in Figure 2 shows AFE-enhanced shadowy “micro-faults,” which is what is being defined.

Since many fractures have little or no throw, their detection is possible by measuring bends in seismic shapes using volume curvature attributes. Volume curvature can detect tight folds at seismic scale that can indicate sub-seismic fractures. While the most frequently used volume curvature attributes are “most-positive” and “most-negative,” the dip curvature attribute can often highlight areas where layers are broken and brittle deformation is present. Volume curvature results are loaded into the AFE process so both actual fracture picks and rose diagram results can be input to the reservoir model.

Seismic attributes can be calculated for surfaces generated from automated 3-D propagation software. Dip, azimuth, and curvature surface attributes are useful for interpreting actual faults and fractures. The surface curvature attribute identifies tight bends just like volume curvature, but for gas shale reservoirs, volume curvature yields better results because reservoirs are up to 1,000 ft thick.

Three-dimensional restoration is an effective way to study stress and strain fields and to better understand the “fracturation” regime. The workflow includes interpreting key horizons, building a geological model, and then using a volumetric restoration tool to unfold the geometries using a geomechanical approach. The resulting paleogeometry and 3-D strain and stress tensors define the internal deformation due to the 3-D restoration.

Wellbore NFS Detection

Cores are the best source for directly measuring and describing natural fracture size, aperture, distance, and orientation. This rich information can add significant confidence to the final reservoir model. Wellbore information such as image logs and sonic logs also can be interpreted to calculate fracture location and orientation. Other well logs can be useful for identifying fractures and their orientation, while providing additional parameters like porosity and permeability which are important in building the reservoir model.

Reservoir Model

Each NFS technique yields different types of information. For example, azimuthal velocities and AVOZ indicate the contemporary stress field; whereas coherency, curvature, dilation, etc. help the geoscientist describe fractures sets related to the current stress or other constraints applied on the field in the past. Results from NFS techniques are correlated and used to improve confidence in the resulting fractured reservoir model.

From a qualitative point of view, these results can be used for field development and well management. When well production comes primarily from the in-situ fracture system, these techniques help identify zones of higher fracture probability. Understanding today’s stress field is crucial to optimizing fracture and front propagation direction during wellbore stimulation or secondary/tertiary recovery.

Modeling and simulation are required for quantitative production predictions. To describe the fractures, all of the information is integrated in a geo-cellular reservoir model to quantify the impact of fractures on reservoir performance. This can be done by generating discrete fracture networks.

Different fracture sets can be modeled separately. Each is simulated by combining its geometric and spatial descriptions (orientation, size, spacing, aperture, etc.) from well information with a related fracture density in a 3-D model. The fracture density information can be derived from many sources: structural (i.e., curvature, distance to fault), geomechanical (strain, dilation from restoration), or geophysical (velocity anisotropy). The effective storage and transport characteristics of the discrete fracture networks can be quantified by upscaling fracture porosity and permeability. These parameters are used to simulate flow in fractured reservoirs and can be calibrated to interpreted well test results.

Though fracture detection is not a straightforward or simple process, flexibility, imagination, and a broad variety of tools and techniques allow even the most complex gas shale fracture systems to be interpreted and successfully exploited using an integrated geoscience and reservoir modeling workflow.