Fracturing Parameter Optimization For Tight Oil Reservoirs Using ANN Surrogate Model

Technological advancement in the exploitation of Tight Oil resources has led to an explosion in crude oil supply and hence crude oil price fall,and fluctuation.However,tight oil production declines fast within the fast years of production,and despite these advancements,the technology is expensive and uncertain hence the need to try and reduce the costs involved while maximizing production.This research aims to determine the primary uncertain factors affecting the oil production in a specific Tight oil reservoir in China and optimize the uncertain fracturing parameters to maximize oil produced and NPV.Inorder to solve this problem data acquired from a specific tight oil reservoir is assessed and used to build a single porosity LGR hydraulic fracture reservoir model and run reservoir simulations by the Schlumberger Eclipse software.Using ANOVA,the factors affecting oil production are determined and ranked.A surrogate model with produced oil and NPV as the objective function is constructed using an artificial neural network in MATLAB,as this limits on the cost and time of individual optimization.Using the developed surrogate mode,GA optimization approach is used in the optimization of the fracturing parameters.After literature study and research on data;8 parameters are identified to have effects on the oil production of tight oil and these parameters include;porosity,permeability,reservoir thickness,reservoir pressure,bottom hole pressure,fracture spacing,fracture half-length,fracture conductivity.It is observed that all these 8 parameters significantly affect oil production during the early stages of production.The most influential parameter overall is permeability.Fracture half-length is the most influential parameter among the fracturing parameters during the early years of production affecting FOPT,followed by fracture conductivity and lastly fracture spacing.Interaction parameters also have a significant influence on FOPT and NPV,and these interaction parameters were mostly permeability by fracture conductivity,permeability by reservoir pressure,and permeability by reservoir porosity.Interactions between only fracturing parameters had no any significant effect.This was outside the expectation which brought to mind the differences of these parameters in affecting oil production for different tight oil reservoirs.The ANN developed surrogate model made optimization easier with lesser computational costs and time in predicting FOPT and NPV.Using GA,the optimal fracturing parameters were in ranges of 60m for fracturing spacing,109m for fracture half-length,and439x10^-3μm²·m for fracture conductivities for both FOPT and NPV at 50USD per bbl oil price for the 3-year period.