Hydraulic fractures from non-uniform perforation cluster spacing: Laboratory study on gelatin and data-driven modeling

Abstract

Hydraulic fracture treatments in horizontal wells involve complexities contemporary analytical and numerical models do not incorporate. Interactions between simultaneously propagating fractures affects neighboring fractures making them grow dissimilarly to each other. This leads to a non-homogeneous fracture geometry, which in many instances results to ineffective or prematurely terminated fractures and therefore sub-optimal economic performance. This project involves the creation of a probabilistic predictive model (PPM), which incorporates empirical inputs from laboratory-scale experimental tests. Random probabilistic distributions are biased statistically in order to generate profiles that match the experimental results. The PPM developed is using Monte Carlo simulation to give a likely final multi-frac geometry.