TY - JOUR
T1 - Surrogate-based optimization of hydraulic fracturing in pre-existing fracture networks
AU - Chen, Mingjie
AU - Sun, Yunwei
AU - Fu, Pengcheng
AU - Carrigan, Charles R.
AU - Lu, Zhiming
AU - Tong, Charles H.
AU - Buscheck, Thomas A.
N1 - Funding Information:
This work was performed under the auspices of the U.S. Department of Energy by Lawrence Livermore National Laboratory (LLNL) under contract DE-AC52-07NA27344 . This work was supported by LDRD-SI program of Lawrence Livermore National Laboratory . We wish to thank Andrew Tompson at LLNL and two anonymous reviewers for their comments that improved the paper.
PY - 2013/8
Y1 - 2013/8
N2 - Hydraulic fracturing has been used widely to stimulate production of oil, natural gas, and geothermal energy in formations with low natural permeability. Numerical optimization of fracture stimulation often requires a large number of evaluations of objective functions and constraints from forward hydraulic fracturing models, which are computationally expensive and even prohibitive in some situations. Moreover, there are a variety of uncertainties associated with the pre-existing fracture distributions and rock mechanical properties, which affect the optimized decisions for hydraulic fracturing. In this study, a surrogate-based approach is developed for efficient optimization of hydraulic fracturing well design in the presence of natural-system uncertainties. The fractal dimension is derived from the simulated fracturing network as the objective for maximizing energy recovery sweep efficiency. The surrogate model, which is constructed using training data from high-fidelity fracturing models for mapping the relationship between uncertain input parameters and the fractal dimension, provides fast approximation of the objective functions and constraints. A suite of surrogate models constructed using different fitting methods is evaluated and validated for fast predictions. Global sensitivity analysis is conducted to gain insights into the impact of the input variables on the output of interest, and further used for parameter screening. The high efficiency of the surrogate-based approach is demonstrated for three optimization scenarios with different and uncertain ambient conditions. Our results suggest the critical importance of considering uncertain pre-existing fracture networks in optimization studies of hydraulic fracturing.
AB - Hydraulic fracturing has been used widely to stimulate production of oil, natural gas, and geothermal energy in formations with low natural permeability. Numerical optimization of fracture stimulation often requires a large number of evaluations of objective functions and constraints from forward hydraulic fracturing models, which are computationally expensive and even prohibitive in some situations. Moreover, there are a variety of uncertainties associated with the pre-existing fracture distributions and rock mechanical properties, which affect the optimized decisions for hydraulic fracturing. In this study, a surrogate-based approach is developed for efficient optimization of hydraulic fracturing well design in the presence of natural-system uncertainties. The fractal dimension is derived from the simulated fracturing network as the objective for maximizing energy recovery sweep efficiency. The surrogate model, which is constructed using training data from high-fidelity fracturing models for mapping the relationship between uncertain input parameters and the fractal dimension, provides fast approximation of the objective functions and constraints. A suite of surrogate models constructed using different fitting methods is evaluated and validated for fast predictions. Global sensitivity analysis is conducted to gain insights into the impact of the input variables on the output of interest, and further used for parameter screening. The high efficiency of the surrogate-based approach is demonstrated for three optimization scenarios with different and uncertain ambient conditions. Our results suggest the critical importance of considering uncertain pre-existing fracture networks in optimization studies of hydraulic fracturing.
KW - Fractal dimension
KW - Global sensitivity
KW - Hydraulic fracturing
KW - Optimization
KW - Surrogate model
UR - http://www.scopus.com/inward/record.url?scp=84879164819&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84879164819&partnerID=8YFLogxK
U2 - 10.1016/j.cageo.2013.05.006
DO - 10.1016/j.cageo.2013.05.006
M3 - Article
AN - SCOPUS:84879164819
SN - 0098-3004
VL - 58
SP - 69
EP - 79
JO - Computers and Geosciences
JF - Computers and Geosciences
ER -