Many studies have assessed microseismic (MS) interpretation in hydraulically fractured shale wells. However, to derive stimulated reservoir volume (SRV) and fracture geometry from MS data is still enigmatic, because MS events come not only from the induced main fractures of the current stage, but also from nonrelevant sources. MS data of adjacent stages tend to overlap each other severely. Simulated hydraulic-fracture (HF) networks that use MS data always yield higher production than actually reported. We addressed these issues by evaluating MS data from the "closure window." With an in-house Excel-VBA program, we divided the MS events from each fracture stage into three windows: the pad window, the proppant window, and the closure window, on the basis of the fracture-stimulation record of each stage. The closure window includes only MS events during the shut-in period (from the end of slurry pumping until before flowback of that stage). During the closure window, leakoff and fracture closing are the dominant phenomena. Leakoff into the formation matrix can cause shear slippage of pre-existing natural fractures (NFs) and tensile opening of micropores. These secondary fractures have potential to transport fluid and to facilitate the induced major fractures. The fracture will close when the fluid inside leaks off. Both leakoff and fracture closing can trigger MS events near the major fractures; thus, they can better capture the effective fracture geometry of the current stage. This method was applied on five shale wells. Unique characteristics in the closure window were observed. The closure window reduced the fracture dimensions of each stage and MS-cloud overlap among stages. Comparing the closure window with the entire window, one finds that fracture width (W-f) decreased the most by an average of 230 ft (26.6%), fracture height (h(f)) decreased by an average of 89.6 ft (25.7%), and fracture length (x(f)) decreased by an average of 41.6 ft (4.9%). The three windows shifted from each other, with closure window shifting from the previous stages. Cumulative-production history-match error dropped from 30% in the entire window to 2% in the closure window. With the closure window, we eliminated those events induced by previous stages, reactivation of NFs, and pumping noise detected by Pad and proppant windows. We suggest that closure window MS data, collected after proppant pumping, provide unique insights into fracture stimulation. The method and VBA program developed in this study can be used to process MS data from any fracture-stimulation job, to segregate the three MS event windows. The MS closure window will indicate fracture geometry more accurately, and thus enhance optimization of hydraulic-fracturing design and the prediction of hydrocarbon production. Infill wells and refracturing may be considered in the light of indicated fracture-geometry reduction.
