Search

Production from two multilateral pads in the Horn River Basin is compared to microseismic-derived parameters. Microseismic was recorded on a near-surface array in both cases. The number of events recorded on each well tends to have a positive correlation to that well's initial production, while the magnitude of those events does not tend to be a good indicator of production in all zones. Fracture models created from located microseismic events also tend to correlate well to production: modeled fracture area, fracture volume and stimulated reservoir volume all show positive correlations. The method in which the rock fractures can also be an indicator of initial production. Wells with higher percentages of dip-slip type rock failures, which can be associated with hydraulic fractures, tend to have higher initial production. In contrast, wells with a larger proportion of strike-slip events, which are typical of fault reactivations in this zone, tend to have diminished production compared to neighboring wells with fewer reactivation events. By understanding what microseismic parameters positively and negatively impact initial production, operators can optimize well production. This can be done in real-time or in installments during lengthy completions programs through the identification of failure type, fracture geometry, and the relative number of events being recorded.

Evaluating hydraulic fracturing treatments by employing passive seismic monitoring technology has become an accepted industry practice and standard in the last decade foremost for unconventional resources. Due to microseismic monitoring it is now understood that the fracture created during hydraulic stimulation greatly deviates from the planar bi-wing textbook example that has been an accepted image in the industry for a long time. In heavily fractured ultra-low permeability shales, the fracture network is highly complex; it is therefore imperative to accurately image this network in order to understand the response of the formation to the treatment as well as the effect of the treatment on production in order to optimize the wellbore completion and the stimulation treatment (Neuhaus et al., 2012). Neuhaus et al. (2012) showcased a multidisciplinary approach to microseismic monitoring by performing an integrated analysis of data acquired during the stimulation of five wellpads completed in the Marcellus Shale. The case study provided a detailed investigation of the microseismic data in conjunction with other data, such as well logs and cores, reservoir properties, and information on regional and local geology. It determined how factors related to the geology of the reservoir and to the stimulation approach impacted the microseismic results. These observations were then used to relate changes in the microseismicity, changes in the geology, and changes in the stimulation method to changes in production in order to optimize field development and the completion design of the wellbores. Recommendations regarding the wellbore azimuth and completion strategy were obtained from the integrated analysis as well as an optimum wellbore spacing. Distinguishing between the total stimulated rock volume (SRV) where microseismic activity was observed and the part of the SRV that contains proppant filled fractures and will therefore be productive in the long term allowed for sophisticated wellbore spacing determination which was performed for one of the pads in the study area. The distribution of proppant filled fractures can also be used to illustrate the containment of proppant within the DFN to evaluate the optimal wellbore spacing, stage length and spacing, as well as landing depth of the wellbore. The microseismic data set used in this case study was acquired with a permanently-installed near-surface array consisting of 101 stations. The wide azimuth, large-aperture, and high fold geometry of the deployed surface array allowed for a consistent resolution under the 18 square mile footprint of the array. Furthermore, the rich sampling of the seismic wavefront provides a high-confidence estimate of event magnitude as well as the failure mechanism for every event which is the crucial input for Discrete Fracture Network (DFN) modelling from microseismic used to evaluate the distribution of proppant and quantify propped half-lengths for the individual treatments (Duncan and Williams-Stroud, 2009; Neuhaus et al., 2012; McKenna and Toohey, 2013). Analyzing the microseismic data in a spatio-temporal sense then enables the reservoir engineer to obtain a system permeability, quantifying the ability to deliver hydrocarbons into the hydraulic fracture system and through the fracture network back to the wellbore, and ultimately predicting production. Overall, the workflow outlined in this extended abstract closes the loop between microseismic data, treatment optimization, and wellbore productivity.

A seismic attribute is a transformation of the original seismic data that can help determine the rock type and fluid characters. AVO attribute analysis is a proven method for confirming seismic amplitude anomalies associated with gas anomalies. AVO analysis examines the intensity of seismic reflections at varying source-receiver distances (offset). The major risk with AVO analysis is whether seismic amplitude anomalies represent commercial or non-commercial gas accumulations. Combining AVO and absorption attributes using ANN reduces this risk. Further risk reduction is accomplished thorough investigation of the different types of absorption attributes and enhancing the ANN training with petrophysical information derived from well logs. We demonstrate that the high frequency content of the seismic response attenuates more as it propagates through gas-bearing reservoirs and, unlike AVO, the absorption attribute is impacted by the amount of gas saturation. Furthermore, we demonstrate how the correlation between shallow gas and shallow seismic amplitudes improves as we include different types of well logs with proper calibration. The ANN, trained by the suite of logs and different frequency-related attributes, enhances the ability to detect undeveloped pockets of shallow gas. Gas has a very marked effect on both density and neutron logs, resulting in lower bulk density and lower apparent neutron porosity. Therefore, combining neutron and density logs and training neural networks based on the well logs with AVO and AQF attributes, instead of just training based on attributes alone, increases the certainty of suspected gas pockets. The real power comes from being able to tie absorption and AVO anomalies with other frequency attributes.