The seismic moment tensor (SMT) is a representation of the strain induced by an earthquake in the vicinity of the hypocenter. In the context of microseismicity associated with treatment of petroleum reservoirs, where events are in response to injection or extraction of materials, the SMT offers the ability observe the instantaneous deformation of the rock mass. For many treatments, the majority of these SMTs are consistent with mechanisms representing the opening and closure of planar cracks; there is a net dilatation or contraction implied for the strain. The natural implication is that the volumetric changes implied by the SMTs are induced by the flow of proppant or other fluids in the system. Therefore, the temporal and spatial patterns of these moment tensors begins to paint a picture of how exactly fluids are moving though the reservoir and how the reservoir is responding to the treatment.
Microseismic monitoring of injection programs in petroleum reservoirs (e.g. hydraulic fractures, cyclic steam injection, and SAGD) as well as carbon sequestration and other scenarios can well-delineate regions of the rock mass that are responding to the treatment. Determination of the polarities and relative amplitudes of the P, SV, and SH waveforms on the geophones that record the event can be used to invert for the SMT (e.g. Trifu et al., 2000), depending if the geometry of the geophone arrays is relatively well-distributed around the seismicity. To be more precise, full determination of the moment tensor requires that there be at least two toolstrings recording the events that are not co-planar with the events themselves (Nolen-Hoeksema and Ruff, 2001). We describe a hydraulic fracture stimulation and cyclic steam production treatment where the geometry of the geophones allowed for robust determination of the SMT for the recorded events. By correlating the type of SMT with the spatial and temporal distribution of the events as well as with the parameters of the treatment, we infer how the medium is responding to the injection or production. We open this paper with a discussion of the theory behind the microseismicity. Then, we present the results of two SMT studies. Contrasting these studies, we show the utility of the moment tensors in determining how the seismicity is developing in relation to the parameters of the injection. Readers who may not be interested in the theory can direct their attention to the examples we present.
A general seismogram is the convolution of a Green’s function, representing how the waves propagate between the source and receiver, with a moment tensor (including the source-time function). The moment tensor is a symmetric, rank-two tensor and is therefore described by six components. In many applications, where it can safely be assumed that the events are pure-shear events occurring on planar faults, only four of these components are used: they can be decomposed into the seismic moment (related to the magnitude of the event); the strike and dip of the fault, and the rake of the motion on the face of the fault.