A New Look At a Famous Well-Analysis of Staged Field Experiment #3 Including Stress Sensitive Permeability in a Multi-Layer Reservoir Simulation Model

摘要

The main issues relating to the performance of hydraulic fractures in tight-gas reservoirs were studied by Holditch beginning in the 1970's and reported in a number of fundamental publications. This work ignored the effect of stress sensitive permeability of the reservoir rock during production and did not model multiple layers with different permeabilities, porosities and water saturations. The authors have recently published on the effect of stress sensitive permeability on welltest analysis results. This work confirmed the conclusion of Holditch that for meaningful analysis of tight gas welltests, numerical simulation is required, rather than welltest analysis. The data set from Staged Field Experiment #3 (SFE#3) provides a test case for the application of these methods. We show that with modern software tools, analysis of post-fracture production data can now be achieved more quickly, with more accurate results, thanks to the addition of stress sensitive permeability and multiple layers in the reservoir simulation model. The SFE #3 fracture treatment data has been re-analyzed with a multi-layer reservoir model for the leakoff from the fracture, while still honoring the original fracture diagnostic data. We constrained the permeability in the fracture model based on the permeability used in the post-fracture history match. A new version of a commercial tool is used to automatically generate a fully functional multi-layer reservoir simulation model, including relative permeability curves for tight-gas reservoirs, stress sensitive permeability for the proppant and the matrix (from published SFE#3 core data). A history match is made of the post-frac water rate and bottomhole pressure by constraining the gas rate, as in the previous work. As in the original work, we also include the effect of proppant pack damage, convergent flow in the fracture and filtrate invasion along the fracture face. A new post-frac production history match is made for SFE#3, with a fracture half-length that now includes the effect of stress sensitive matrix permeability and a more realistic multi-layer reservoir model. The results are consistent with the fracture modeling and fracture diagnostic results previously reported. A better understanding of the reasons for short "apparent" fracture lengths seen in conventional welltest analysis is obtained by also including the effect of stress sensitive matrix permeability in the history match of the benchmark SFE#3 data set, using core data obtained in the original project. The effect of using a more realistic multi-layer reservoir model is also shown. The use of modern tools is demonstrated which makes this workflow practical for routine analysis of tight gas wells, rather than something only to be used for research, as in the past. Finally, the new model is used to predict the long-term production from the well, and this value is compared to the original prediction from Holditch's single well model and the actual production from the well, obtained from public data. The comparison with the actual 15 year production history yields valuable insights into the effective drainage area, GIIP, recovery and long-term water production mechanism.