The efficiency of hydraulic fractures largely determines the economics of tight sands and gas shale reservoirs. Efficiency is defined by the fracture conductivity which results from the stimulation treatment. This is a function of fracture half length and height. Independent measures of fracture length and height, and containment provide prime data in evaluating stimulation treatments. Downhole or surface microseismic monitoring produce spatial distributions of microseismic events associated with hydraulic fracturing. These distributions provide the constraints on fracture length, height and containment. Most commercial product offerings ignore uncertainty in locations and biases produced by limited observation arrays. We examine these uncertainties and biases through an analysis of a simulation of event locations and sensor arrays. In addition, we investigate the errors produced by ignoring velocity anisotropy characteristic of gas shale.
We studied a set of synthetic microseisms locations. Simulations results demonstrate that mapping fracturing geometry with subsurface sensor arrays requires at least two observation wells (not a common field practice); this minimizes the geometry bias generated when only one observation well is used. One of the two observation wells should be aligned in the direction of the fracture azimuth. Failure to do so results in underestimation of the fracture half-length by approximately 10%. This requires knowledge of the prevailing in-situ stresses directions prior to stimulation. Downhole sensors should be located at the same depth of the fractured formation to avoid errors in fracture height estimation which can be as much as 50%. Velocity anisotropy has a minimal influence on uncertainty in event locations when subsurface sensor arrays are modeled with horizontal velocities. Surface sensor arrays can lead to underestimation of the fracture half-length in the order of 30% to 40% depending on the anisotropy effect whereas fracture height estimation error can be as much as 4%. Analysis of surface sensors arrays which avoid production shutdowns, reveals fracture half-lengths can be underestimated by about 17% to 31% depending on the velocity anisotropy. Uncertainty in event locations using only the vertical velocity in an anisotropic velocity field reduces with increasing depth as expected. Recent analysis of spatial distributions of microseismic activity for effective permeability and diffusion coefficients requires improved location accuracy and estimations of uncertainty.