Third EAGE Passive Seismic Workshop - Actively Passive 2011

The localization of seismic events is of great importance not only in seismology but also in exploration geophysics for monitoring of for instance hydraulic fracturing. It can be successfully implemented by diffraction stacking, where the source location is obtained from the maximum of the image function. Since the maximum of the image function is distinct, even very weak events can be detected. Previous research showed that the method works perfectly for homogeneous 2D media; we demonstrate the extension to 3D. Numerical examples in both homogeneous and heterogeneous media are presented. Strongly heterogeneous media are intensely affected by triplications. Localization of such events can be proceeded by taking the most energetic events into account. Moreover, by using geometrical spreading as weighting factors for the input data, separation of the propagation and source effects is achieved. Also effects of the double couple radiation pattern were investigated. Furthermore, the method was tested on field data from Southern California. Before applying diffraction stacking it was accounted for the acquisition footprint. The localization results deviate from the source position obtained by a standard picking method less than 1 km in all three directions. Both numerical and field data applications reveal the potential of the method.

Microseismic monitoring is synonymous with shale gas development in North America: experience with shale plays suggests a high degree of variability in rock properties, subsequent completion effectiveness, and ultimately gas production and EUR. There are several modes of acquisition currently in use to obtain microseismic information which is related to the hydraulic fracturing process. The study reported here focuses on borehole microseismic and attempts to understand the information content in ‘conventional’ monitoring practice available in the industry. It examines the sensitivity to velocity model and uncertainty in derived data and attributes. Four stimulations and two monitor wells provide 12 multi-stage datasets for repeated analysis. Monitoring with a single well tends to produce bias in results towards the observation. Further, we see that a horizontal monitor provides different controls on locations to a vertical monitor, but their combination in a dual-well solution usually produces the more consistent view. As we change the velocity model, we see more variation in located events for the same input data (extent in 3D and possible orientations). The interpretation of these results in terms of hydraulic fracture characterization and effectiveness is therefore non-unique, and this uncertainty should be recognised.

In this study we examine the role of various types of failures associated with a tight sand hydraulic fracture treatment, utilizing moment tensor inversion and source parameter analysis of recorded microsiesmic events. Based on nearest neighbor statistics, events are grouped into near treatment well and fracture extension regions, and used to outline a discrete fracture network (DFN) and the spatial-temporal development of the DFN within the volumes. These results are further used to establish a geomechanical model to assess the permeability associated with the treatment, which is further used to estimate the Stimulated Reservoir Volume (SRV) associated with the treatment program and possibly enhance calculations for productivity.

We acquired at Peace River a suite of geophysical data including 3 months of passive seismic. A surface array of buried multi-component receivers covering an area of about 2Km^2 listened to the recovery activity, which is driven by steam injection at a depth around 550m below the surface, with the intention of detecting induced microseismic events. To support this information, a series of string shots were detonated in a well near the center of the survey area and recorded by the same receiver array. We benchmarked the localization results of two service companies against an in-house quality control processing based on a cross-correlation migration approach. From the analysis, we found that: (1) the minimum detectable moment magnitude by the surface array seems to be around –1.6; (2) the catalogues provided by the service companies differ greatly, and contain a significant number of events that appear not to be associated with the activity ongoing at the reservoir; (3) the events categorized as microseisms are located near the heel of a well that was injecting steam at the time of the survey; (4) the differences in location of string shots and microseismic events between our and the companies’ processing are less than 100m.

Passive seismic methods which use no active human-build source, but utilise the natural sounds of the earth, produce in most cases a low resolution of the subsurface compared to active seismic. Due to the usually low-frequency content of the earth’s seismic sources, they are often dismissed and not even considered as an add-on for an active-seismic campaign. And this, despite a similar resolution than gravimetric methods, minimal costs compared to an active survey, the same “domain” as active seismics, and their environmental friendliness.

Microseismic monitoring can be either carried from the surface or borehole. While noise levels in the borehole are usually much lower than on the surface (typically by a factor of 10), we show that the signal amplitude at the earth’s surface is also larger due to a free surface boundary condition (simultaneous observation of both direct and reflected waves) and impedance contrasts at the low-impedance near surface layers. Thus a comparison of detectability of microseismic events between surface and downhole monitoring systems needs to take into consideration amplitude gains in near surface layers. We show a typical example of such gains on a vertical geophone array where the amplitude of the direct P-wave increases by more than a factor of 3 from a 150 m deep geophone to a geophone at the earth’s surface. Considering this amplitude gain one can calculate a theoretical detectability between surface and downhole monitoring arrays and we show that surface and downhole monitoring arrays can have approximately similar detectability at distances of a few hundred meters from a monitoring borehole, and that a surface array detects significantly more events at distances beyond 500 meters from a monitoring borehole.

The analysis of passively recorded low-frequency microtremors for detecting hydrocarbon (HC) reservoirs is an emerging technology, which has created some debate in the last few years. On one hand, these microtremors have been proposed as a reservoir indicator with potential for optimizing well placement during exploration, appraisal, and development. On the other, pitfalls in both the analysis of low frequency passive seismic and the proposed models of its generation have been also reported when anthropogenic noise overlaps and is mistaken for possible HC microtremor. During April/May 2010 a low frequency passive seismic survey was undertaken over an exploration area in the Albertine Graben, Uganda, to record seismic tremor and to explore the possibility of its exploitation in this area. In this paper, we describe the survey and resulting analysis of data recorded at locations above and away from the reservoir, and we investigate the temporal and spatial distribution of the data spectral content. A definite spatial distribution of derived seismic wavefield attributes exists and it is proposed that these observations are more likely to be related to the geology and structure of the sedimentary basin rather than to the presence of HC.

We have developed a new multilevel seismic downhole array for the purpose of ambient seismic noise measurements. The tool substantially widens the bandwidth and lowers the detection threshold for microseismic events at low signal-to-noise levels near the ambient seismic noise floor. We analyse the bandwidth requirements in both frequency and dynamic range needed for accurate determination of dynamic source properties of microseismic events, such as the seismic moment tensor and event stress drop and compare them with the designed capabilities of this new tool.

PAS21