A semi-analytical flowback data history-matching approach that rigorously accounts for fracture dynamics

Abstract

Flowback data analysis methods used for unconventional reservoirs are still in their infancy. Until now, the dynamics of fracture closure during flowback have not been rigorously accounted for in analytical or semi-analytical models. In some cases, at the start of flowback, fracture pressure may be higher than the minimum horizontal stress, which means the fracture is still closing during the early flowback period. For such cases, the commonly-applied assumption of a static fracture width in flowback modeling will be in significant error. Moreover, previous analytical and semi-analytical modeling approaches have ignored fluid leakoff from the fracture when the fracture pressure is greater than reservoir pressure, which could also lead to significant errors when analyzing early flowback data in a stress-sensitive reservoir.In the current work, a semi-analytical history-matching approach for characterizing hydraulic fractures, which captures the dynamic behavior of hydraulic fractures during the early production period, is presented. The proposed methodology accounts for varying fracture width, cumulative leakoff from the fracture to the matrix, and spontaneous imbibition, during early flowback. The new method is verified using numerically simulated data and its practical application demonstrated by history matching a field case (shale gas well) exhibiting evidence of dynamic fracture properties during flowback. History matching of the same field case using models that do not account for fracture dynamics or fluid leakoff reveals that these models may overestimate fracture properties such as initial fracture permeability. In addition, history matching of the field case with the new model without spontaneous imbibition effects implemented yields fracture properties (initial fracture permeability and fracture half-length) that are greater than when both geomechanics and spontaneous imbibition are implemented.When compared to existing flowback models, the proposed semi-analytical model may yield more accurate estimates of fracture properties when fracture dynamics cannot be ignored during the early flowback period.