DE-FC26-06NT42956

Goal
The objective of this project is to locate and characterize natural fracture “sweet spots” and induced (hydraulic) fractures from scattered wave analysis of 4-D vertical seismic profile (VSP) data in order to optimize well placement and well spacing in low-permeability gas reservoirs.

Performer
Massachusetts Institute of Technology, Cambridge, MA

Results
The first year of the project has focused on numerical modeling and the development of processing workflows for identifying and analyzing scattered wave signals from hydraulic fractures. Researchers have adapted and tested a 3-D finite difference model to generate elastic VSP and microseismic data for a simplified hydraulic fracture in a layered earth. Synthetic 3-D VSP data were generated for an initial range of fracture parameters, including two different fracture compliance values. The results illustrate the sensitivity of the scattering amplitude to fracture compliance (aperture). Researchers have completed extensive simulations of VSP data sets for unfractured, compliant, and stiff fracture models, developed processing software for component rotation, arrival picking, time-differencing, and stacking of synthetic 4D VSPs and microseismic events, and tested a methodology for extraction of P and S scattered wave energy from 4D VSP synthetics for the characterization of fracture properties (compliance).

Three sets of model runs were completed. The first set was for a single fracture in a simple layered background model. The single fracture model was used to test the effectiveness of using the microseismic event moveouts as stacking operators for P and S scattered wave signals. The resulting fracture diagnostics indicate that open fractures (those with high compliance) have larger P and S wave scattered amplitudes than closed fractures (those with low compliance). These results provide confidence that the basic methodology can isolate the fracture signals and differentiate between open and closed fractures.

A second set of models was run with two fractures in the subsurface. The purpose of this modeling effort was to determine if it was possible to separate the scattered wave signals from different fractures. The presence of multiple fractures results in more complex scattered wave signals, making it more difficult to isolate and analyze the signals from a single fracture. For this test, the two fractures were assumed to be in different wells, meaning that they were separated spatially. By applying the P and S moveouts obtained from microseismic events at each fracture location, we were able to isolate the scattered wave VSP data for analysis.

During the 4th quarter we completed our modeling effort by creating a velocity model of the field site based on acoustic log data. Using this new velocity model, we then simulated the effect of three fractures. The three fractures were distributed to mimic the geometry of the field fracture program. Specifically, two fractures were placed directly above one another in the same treatment well with a separation of 100m, while the third fracture was placed in a different well but at the same depth as the deep fracture in the first well. This final modeling step was done to ensure that the individual fracture signals could be identified in a geometry similar to the field site

The results indicate that the scattered energy is greatest, and therefore most easily detected, on the records of the acquisition geometries where the sources and receivers are inline with the normal to the fracture surface. For the stiff fracture it is difficult to see the scattered energy on the other, non-normal, raw VSP cases. For the compliant fracture case, the scattered energy is observable on nearly all the raw VSP records. Depending on the actual fracture compliances in a reservoir, it is likely that we will be able to detect this energy. Following the extraction of the time-lapse signals due to fractures, our method of using event moveouts extracted from microseismic data appears to do a good job of flattening the VSP records and providing a measure of fracture scattering as a function of the seismic illumination direction. Initial results from these synthetic data also suggest that it may be possible to discriminate between open and closed fractures by the azimuthal change in the scattering as seen on the P and S wave RMS energy ratio plots. The modelling results show that it may be possible to extract the relative quality of induced fractures within a tight gas reservoir from a time lapse VSP when used in conjunction with microseismic recordings. The workflows developed in this modeling phase will be applied to data from a Wyoming tight gas field in Phase II of the project.

Benefits
By using time-lapse VSP imaging before and after hydraulic fracturing, researchers will develop methods for characterizing the flow properties of fractures from seismic data. An improved understanding of the geometry and flow characteristics of hydrofracs will allow producers to optimize the number and location of wells and fractures. Hydrofracture compliance, which is related to fracture aperture, will be characterized by developing a diagnostic from the seismic scattered wave amplitudes. The seismic scattering signatures of these “known” hydraulic fractures will also provide a much-needed calibration point for identifying natural fracture sweet spots from both surface seismic and VSP data.

Background
Tight gas sand reservoirs generally contain thick gas-charged intervals that often have low porosity and very low permeability. Natural and induced fractures provide the only means of production. If sweet spots containing natural fractures can be identified and located, then productive, economic wells can be brought on line. Likewise, if induced (i.e., hydraulic) fractures can be strategically placed to both connect with existing natural fractures and avoid interaction with reservoir sections being drained by other wells, then the ultimate recovery from these reservoirs can be maximized.

Summary
During the course of this first year we have modeled the seismic response of single and multiple hydraulic fractures using a 3-D elastic finite difference approach. Using the results of the single fracture model we developed a methodology or workflow for isolating the scattered wave energy from the VSP records and stacking the P and S wave events using operators extracted from the microseismic records. In summary, 3-D VSP data acquired before the hydraulic fracturing is subtracted from the same data acquired after fracturing, resulting in a time-lapse (4-D) VSP data volume. This data volume contains scattered wave signals generated by the interaction of the propagating seismic energy and the hydraulic fracture. Microseismic events generated by the fracturing process, and recorded in the same monitoring well used for VSP acquisition, provide a direct measure of the travel paths of data from the fracture to the receivers. The moveout of P and S wave arrivals from the microseismic events across the receiver array can be used as an operator for stacking the 4-D VSP scattered wave signals. The results are P and S scattered wave amplitudes as a function of illumination angle or azimuth for each surface source point in the VSP. These amplitude values provide a diagnostic of fracture quality based on the fact that high compliance ‘open’ fractures scatter more seismic energy than low compliance ‘closed’ fractures. While it is impossible to characterize the rock volume associated with each microseismic event, we expect to compare the relative scattering for small rock volumes, perhaps as small as 50m on a side, across the reservoir. While the actual properties of the fractured zone itself may prove difficult to estimate, we should be able to map the areas with relatively higher fracture quality, i.e. open and compliant properties. The right column of the Figure shows that there is about a factor of five between the amount of scattered energy measured for the compliant and stiff fracture cases. This means that the economically important, open and compliant fractures will scatter more energy than the less important, closed and stiff fractures. It is much more likely that the fractures contributing to permeability of the gas will be the ones that will be detectable by this method. It may also be possible to discriminate between open and closed fractures by the azimuthal change in scattering as seen on the P and S wave RMS energy ratio plots, although the measurement errors may not allow this level of quantification.

Current Status (February 2008)
The project has completed its first year of activity. The first phase focused on numerical simulation of seismic scattering from hydraulic fractures and the development of processing and analysis methodologies to extract fracture quality diagnostics. The second phase of the project will focus on analysis of field data from a tight gas sand reservoir in Wyoming.

Funding
This project was selected in response to the DOE National Energy Technology Laboratory Fossil Energy Research and Development solicitation DE-PS26-06NT42787.

Project Start: October 1, 2006
Project End: September 30, 2008

Anticipated DOE Contribution: $579,738
Performer Contribution: $395,526 (40% of total)

Contact Information:
NETL – Chandra Nautiyal (chandra.nautiyal@netl.doe.gov)
MIT – M. Nafi Toksoz (toksoz@mit.edu or 617-253-7852)

Polar plots of the average scattered P wave energy (top row) and S wave energy (bottom row) as a function of angle of incidence of seismic energy on the fracture. The compliant and stiff fractures results are shown in columns one and two, respectively. Column three shows the ratio of the Compliant to Stiff results. Note the 0 degrees azimuth denotes normal to the fracture strike.