Tracer Flowback Based Fracture Network Characterization in Hydraulic Fracturing

Abstract

Abstract Horizontal multi-stage fracturing has been currently the primary method for developing tight reservoirs. However, the unavailability of quantitative fracture network characterization remains an impediment in this regard. In light of this, this present work aims at characterizing the fracture morphology of each fracturing stage by using tracer breakthrough curve (BTC) from fracturing fluid recovery. This shall provide a solid guidance for the quantitative recognition of discrete fracture network in tight reservoirs development. Based on field-scale tracer application in Xinjiang oilfield, fracture networks mapped through microseismic monitoring were firstly categorized upon fracture scales. Then correlative relationship between tracer BTCs and fracture morphology categories was built through statistics approach. A single-well tracer injection-withdrawal seepage model which permits an explicit modeling of discrete fractures was developed based on discrete fracture model (DFM) to perform history matching of field tracer BTC. Finally, sensitivity analysis was conducted to quantitatively uncover the general influence of fractures statistical parameters, i.e. central point density, mean and standard deviation of orientation, length and aperture, on the initial concentration, peak concentration and signal duration of BTC. According to microseismic monitoring results, fracture network can be roughly categorized into micro fractures, large fractures and their mix. Corresponding tracer BTCs for different fracture network categories indicate: when micro fractures dominate, tracer BTC gives a normal distribution type with a lower initial and peak concentration and a longer signal duration; when essentially large fractures present, tracer BTC behaves as a unimodal type with a higher initial and peak concentration and a shorter signal duration; when large and micro fractures coexist, tracer BTC resembles a parabolic unimodal type with a higher initial concentration and a longer signal duration. Sensitivity studies based on DFM simulation show: (1) the statistical parameter of mean has a more profound impact on tracer BTC than that of standard deviation; (2) a higher fracture central point density contributes to a larger fracturing range, lower initial and peak concentration and a longer signal duration; (3) fracture orientation exerts a determinant role in the number of fracture intersections with horizontal wellbore, and a higher initial and peak concentration occurs when fractures are approximately parallel to the wellbore; (4) a shorter fracture length will lead to a higher initial and peak concentration due to the limited tracer propagation area underground; (5) with the increase of fracture aperture, more serious flow-filed dispersion happens and thus lower initial and peak concentration occurs. A correlative relationship between fracture network spacial configuration and tracer BTC in tight reservoirs has been obtained via statistically analyzing tracer BTCs for different fracture network types. Furthermore, sensitivity analysis helped single out the key controlling factors for quantitatively recognizing complex fracture networks, which will offer practical means of characterizing discrete fractures for fracturing effect assessment and dynamics prediction.