An integrated study on the sensitivity and uncertainty associated with the evaluation of stimulated reservoir volume (SRV)

Abstract

Economic production from ultra-tight shale formations depends greatly on the effectiveness of the hydraulic fracturing stimulation treatment. As a critical monitoring technology for unconventional reservoir development, microseismic plays an important role in fracture location analysis and evaluation of the fracturing treatment. For microseismic analysis, particularly for fracture inversion, uncertainties are common and are associated with the collected microseismic data themselves and formation properties such as velocity analysis result. Therefore, large differences are often observed between the fitted stimulated reservoir volume (SRV) from microseismic events and the volume obtained through hydraulic fracture modeling.This paper combines the microseismic monitoring, hydraulic fracturing simulation and SRV fitting techniques to investigate the inter-relationship of data uncertainties and to provide a possible methodology for guiding the stimulation treatment during reservoir development. The hydraulic fracturing process is simulated to obtain a realistic set of microseismic events that are generated from hydraulic fracture propagation and the pore pressure increase in the matrix. In order to examine the impacts of the uncertainties on the estimated locations of the microseismic events, different levels of noise are added to the first arrival times during the inversion process. Because the microseismic event locations inverted from the first arrival times with noise are very scattered, a denoising procedure based on probability theory is applied to identify and exclude the outliers in the MS event clouds. The discrete bin, 3D Delaunay triangulation and ellipsoid fitting methods are used to compute SRV from the inverted microseismic events with noise. Comparisons are made to examine the sensitivities of the different SRV fitting methods with respect to data noise. The results show that the two discrete bin algorithms are relatively stable, while the MVEE and 3D Delaunay triangulation algorithms are not very robust and can be easily affected by the noise in the observed data.