Automatic fracture optimization for shale gas reservoirs based on gradient descent method and reservoir simulation

In shale gas reservoir development, determination of hydraulic fracture geometry for horizontal wells is a demanding yet challenging task. One type of approach for hydraulic fracture optimization is based on reservoir simulation. To improve optimization efficiency and accuracy, an automatic and robust procedure integrating the gradient descent method with gas reservoir simulation has been developed. Fractured reservoir models were constructed using the "Multiple INteracting Continua" method, whereby an in-house shale gas reservoir simulator was implemented to model multiple gas transport mechanisms including non-Darcy flow, gas desorption, Klinkenberg effect, and geomechanical effect. The optimization procedure was first validated against two ideal cases and then applied to two realistic cases to optimize fracture spacing, half-length, and dimensionless fracture conductivity. It showed that the optimization results depend on optimization objective, reservoir property, natural fractures, economics and termination criteria. This gradient descent assisted fracture optimization procedure can achieve significant computational reduction and high prediction accuracy for various shale gas reservoir cases.