Hydraulic fracturing is considered as one of the most effective stimulation techniques to improve recovery especially from unconventional low permeability reservoirs. However this promising stimulation technique sometimes does not respond as expected. Significant amount of work has been dedicated to this topic with ineffective fracturing fluid (FF) clean-up considered as one of the main reasons for this underperformance. However there are still great deals of uncertainties in this area primarily due to large number of parameters affecting FF invasion and its back flow. This work presents results of 10 different sets of numerical simulations consisting of injection, soaking and production periods for 40960 runs. Each set consists of 4096 runs and investigates the simultaneous impact of 12 pertinent parameters (fracture permeability, matrix permeability (km), end points and exponents of Corey gas and FF relative permeability curves in both matrix and fracture and matrix capillary pressure (Pcm) (depending on interfacial tension (IFT), km and pore size index). Two-level full factorial experimental design and linear response surface statistical approaches were used to sample the variables domain, covering a wide practical range determined with the support of our 11 industrial sponsors, and generate output response. Results indicate that improvement in FF mobility inside the fracture is the major factor affecting FF cleanup efficiency. In line with this finding, maintaining high Pcm, by retaining high IFT, results in cleaner fracture (lower gas production loss, GPL). That is, increasing IFT retains FF within the matrix and allows more gas to flow freely inside the fracture. This was confirmed by the corresponding saturation map of FF distribution. The effect of Pcm was more pronounced when drawdown was very low and/or soaking time was extended. At very low drawdown and when km was reduced, in a set within its variation range, the effect of the resultant increase of Pcm on GPL was more pronounced than that of the resultant decrease in FF mobility. Generally when injected FF volume increased, larger GPL was observed and reduction of GPL (cleanup) was also slower. As fracture length decreased, cleanup was faster and the effect of fracture pertinent parameters on GPL, compared to those of matrix, decreased. This works findings allows better evaluation of benefits of this costly operation leading to an optimized design and more efficient ways to improve its performance. For instance, sometimes aiming for a longer fracture, due to its FF poor cleanup performance, is not practically attractive and use of IFT reducing agents to produce more FF during the back flow period, would not have the intended impact to bring back its performance to the desired ideal level.

No abstract

No abstract