Use of Extremely High Time-Resolution Production Data To Characterize Hydraulic Fracture Properties

Abstract

Abstract By employing flowback surface rate and pressure data recorded at very high frequency immediately following hydraulic fracture stimulation shutdown, it is possible to obtain estimates of fracture conductivity, fracture length and near-wellbore matrix permeability, as well as confirmation of closure pressure. Because the wellbore and induced hydraulic fracture are liquid filled at very early time, it is possible to observe meaningful signatures in the flowback production data when it is recorded with sufficiently high time resolution, on the order of 20 or more observations per minute. Dissipation of the super-charge is detectable, and the highly elliptical flow dominated by the fracture is clearly evident. Recognition of these flow regimes makes interpretation of fracture conductivity, effective length and an upper limit estimate of matrix permeability possible. Examples of actual field results are shown in the paper. When compared with published proppant bed conductivities for the expected stress loading, excellent correlation is obtained. The computed effective fracture lengths are somewhat shorter than that expected based on the design. In all cases, the near-fracture matrix permeabilities are higher than observed in later time production histories. Introduction The primary value being sought in hydraulic fracture design and execution is fracture conductivity. Without conductivity, the effective fracture length will always be short. Fracture assessment techniques (1–10) have provided a foundation to demonstrate that characterization of the fracture is feasible. With sufficiently frequent rate and pressure observations (11), characterization of the fracture and near-wellbore properties is relatively straight forward. This paper demonstrates a method for evaluating the proppant bed, fracture length and near-wellbore properties. An example is provided which demonstrates the impact that conductivity has on ultimate effective fracture length. The method to be employed is the Reciprocal Productivity IndexTMmethod (11–13). By virtue of the theoretical basis provided by the method for very early times (11) and affirmed by the theoretical development of Carslaw and Jaeger (14), the early time production data including all produced phases (water, gas, oil or condensate and proppant, if significant) and their flowing pressures are interpretable, once corrected to sandface conditions. Methodology Historically, production histories from post-stimulation flowbacks have not been viewed as amenable to quantitative interpretation. One of the earliest documentations that the data was useful (7) showed that meaningful fracture and reservoir information could be extracted. Since then several authors have extended that work (3–6, 11). In order to have an interpretable dataset, all the phases which contribute to the consumption of energy in the reservoir and fracture must be measured frequently and early enough that the pressure transient is predominantly moving in the fracture. That means that while in some systems hourly data may be sufficient, generally, multiple observations per minute are required. Conventional wisdom has held that wellbore storage masks the phenomena of interest. However, since prior to gas breakthrough in the fracture, the system is single phase liquid for most stimulations, the controlling physics are those of a transient passing through a liquid-filled "vertical pipeline" such as treated by Wylie and Streeter (15). They show that transit times are in the order of tens of seconds for wellbore / fracture systems, regardless of flow rate. Data collection at that time resolution is now quite feasible.