The treatment of diatomite reservoirs with cyclic steam is intended to enhance the mobility of the oil in reservoir, thereby stimulating hydrocarbon production. With optimized array geometries, the advanced analysis of the microseismicity resulting from these treatments can yield much more information on these events, such as their failure mechanisms through the process of moment tensor inversion. When these mechanisms suggest that the strain induced by the events is occurring on natural or newly created fractures, their orientations may be gleaned from the moment tensor. In this study, we monitor microseismic events observed during the injection and production stages of a ‘Huff and Puff’ steaming operation in a diatomite. Our focus is on analyzing the growth of two event clusters that occurred during production cycles after steaming. Since the events are located using three downhole, we can invert for the components of failure. One cluster showed significant vertical growth and the mechanisms of the microseisms were consistent with dilatational opening of fractures, closures of the fractures, or double-couple (shear) events. While the trend of the cluster of events is reflected in the fracture orientations of some of the events, many of the events have fractures occurring at defined angles to the overall trend. This observation allowed us to infer that the stress regime under which most of the events were responding is dynamically changing as the strain induced by surrounding events evolves in the reservoir. The second grouping of events was spatially contained and consisted of inflationary or deflationary failure mechanisms, suggesting containment that reflects steam chamber development; further exemplified by the variability in fracture azimuths, suggesting that the local stress conditions played a more significant role in the observed failures. Based on this study, we suggest that advanced microseismic analysis has the potential to outline when steam chamber development occurs within design specifications and when the regional/local stress conditions influence steam chamber development, such as observed in the example of poor containment. In both examples, the occurrence of events during production suggests that failures occur as a result primarily as a result of stress unloading and stress re-distribution locally.

Three monitoring wells with permanently installed 3C geophones were used to locate the microseismicity induced from a steam production cycle. This dataset of 52 high signal-to-noise ratio events were used to examine the location of events by progressively decimating the number sensor arrays. The three-array solutions were contrasted with different array combinations achieved by turning off one or two of the arrays. Event locations revealed the nature and magnitude of the limitations having incomplete coverage of the treatment zone. Most steam and fracture treatments are monitored by a single observation well. Parameters, such as stimulated reservoir volume fracture azimuth and fracture dimensions in treatments like CSS, hydraulic fracturing, SAG-D, are estimated from the distribution of microseismic event locations. By taking three array locations as ground truth, the array configurations that most accurately reflect the actual fracture geometry are determined. The observed distributions of the events relocated with decimated arrays show significant changes in overall fracture trend, geometry, and location accuracy. Progressive decimation of the number of arrays increases the inaccuracies of event locations, which results in the scatter of event locations. Decimation of the number of arrays changes the dimensions and azimuth of fractures, which are readily apparent upon comparison with the three-array solutions. The least biased decimated solutions are those using two arrays, one on either side of the treatment zone. Single array solutions show the most scatter, with the recording distance also controlling the degree of event mislocation. This array decimation analysis shows how the array configuration and number for arrays affect the interpreted fracture volumes, geometries, and accuracy of event location. The different viewing angles from the single and dual array decimated subsets, as well as the different distances from the event clusters to the geophones are important considerations in mitigating the biases due to limiting the recording coverage.