Abstract: A method for history matching a reservoir model based on actual production data from the reservoir over time generates an ensemble of reservoir models using geological data representing petrophysical properties of a subterranean reservoir. Production data corresponding to a particular time instance is acquired from the subterranean reservoir. Normal score transformation is performed on the ensemble and on the acquired production data to transform respective original distributions into normal distributions. The generated ensemble is updated based on the transformed acquired production data using an ensemble Kalman filter (EnKF). The updated generated ensemble and the transformed acquired production data are transformed back to respective original distributions. Future reservoir behavior is predicted based on the updated ensemble.

Abstract: A system is described for estimating well production and injection rates of a subterranean reservoir using machine learning models. The system may include a processor and a non-transitory computer-readable medium comprising instructions that are executable by the processor to cause the processor to perform various operations. The processor may receive a set of static geological data about at least one subterranean reservoir in a subterranean formation. The processor may apply a trained convolutional neural network to the set of static geological data and data on initial states of dynamic reservoir properties to determine dynamic outputs of the subterranean reservoir. The processor may determine well data by extracting the set of static geological data and the dynamic outputs at well trajectories. And, the processor may apply a trained artificial neural network to the well data and subterranean grid information about the subterranean reservoir to generate estimated well production and injection rates.

Abstract: A system and method for controlling a gas supply to provide gas lift for a production wellbore makes use of Bayesian optimization. A computing device controls a gas supply to inject gas into one or more wellbores. The computing device receives reservoir data associated with a subterranean reservoir to be penetrated by the wellbores and can simulate production using the reservoir data and using a physics-based or machine learning or hybrid physics-based machine learning model for the subterranean reservoir. The production simulation can provide production data. A Bayesian optimization of an objective function of the production data subject to any gas injection constraints can be performed to produce gas lift parameters. The gas lift parameters can be applied to the gas supply to control the injection of gas into the wellbore or wellbores.

Abstract: A system is described for estimating well production and injection rates of a subterranean reservoir using machine learning models. The system may include a processor and a non-transitory computer-readable medium comprising instructions that are executable by the processor to cause the processor to perform various operations. The processor may receive a set of static geological data about at least one subterranean reservoir in a subterranean formation. The processor may apply a trained convolutional neural network to the set of static geological data and data on initial states of dynamic reservoir properties to determine dynamic outputs of the subterranean reservoir. The processor may determine well data by extracting the set of static geological data and the dynamic outputs at well trajectories. And, the processor may apply a trained artificial neural network to the well data and subterranean grid information about the subterranean reservoir to generate estimated well production and injection rates.

Abstract: A system and method for controlling a gas supply to provide gas lift for a production wellbore makes use of Bayesian optimization. A computing device controls a gas supply to inject gas into one or more wellbores. The computing device receives reservoir data associated with a subterranean reservoir to be penetrated by the wellbores and can simulate production using the reservoir data and using a physics-based or machine learning or hybrid physics-based machine learning model for the subterranean reservoir. The production simulation can provide production data. A Bayesian optimization of an objective function of the production data subject to any gas injection constraints can be performed to produce gas lift parameters. The gas lift parameters can be applied to the gas supply to control the injection of gas into the wellbore or wellbores.

Abstract: A history-matched oilfield model that facilitates well system operations for an oilfield is generated using a Bayesian optimization of adjustable parameters based on an entire production history. The Bayesian optimization process includes stochastic modifications to the adjustable parameters based on a prior probability distribution for each parameter and a model error generated using historical production measurement values and corresponding model prediction values for various sets of test parameters.

Abstract: Certain aspects and features relate to a system that projects an optimized foam-fluid-application scenario for to stimulate production from a hydrocarbon well. The optimized scenario can include a recommended chemical make-up for the foam entity as well as application parameters such those related to timing and duration. A hybrid discrete fracture network and multi-porosity (DFN-MP) model for fluid interaction with the formation where the well is located can be produced by a processing device. The hybrid DFN-MP model can be optimized using field simulation data for the formation. The optimized hybrid DFN-MP model can be used by the processing device to produce an optimized foam-fluid-application scenario, which can be communicated to at least one well and can be utilized to stimulate the well for increased production.

Abstract: A method for history matching a reservoir model based on actual production data from the reservoir over time generates an ensemble of reservoir models using geological data representing petrophysical properties of a subterranean reservoir. Production data corresponding to a particular time instance is acquired from the subterranean reservoir. Normal score transformation is performed on the ensemble and on the acquired production data to transform respective original distributions into normal distributions. The generated ensemble is updated based on the transformed acquired production data using an ensemble Kalman filter (EnKF). The updated generated ensemble and the transformed acquired production data are transformed back to respective original distributions. Future reservoir behavior is predicted based on the updated ensemble.

Abstract: Using production data and a production flow record based on the production data, a deep neural network (DNN) is trained to model a proxy flow simulation of a reservoir. The proxy flow simulation of the reservoir is performed, using an ensemble Kalman filter (EnKF), based on the trained DNN. The EnKF assimilates new data through updating a current ensemble to obtain history matching by minimizing a difference between a predicted production output from the proxy flow simulation and measured production data from a field. Using the updated current ensemble, a second proxy flow simulation of the reservoir is performed based on the trained DNN. The assimilating and the performing are repeated while new data is available for assimilating. Predicted behavior of the reservoir is determined based on the proxy flow simulation of the reservoir. An indication of the predicted behavior is provided to facilitate production of fluids from the reservoir.

Abstract: Certain aspects and features relate to a system that projects an optimized foam-fluid-application scenario for to stimulate production from a hydrocarbon well. The optimized scenario can include a recommended chemical make-up for the foam entity as well as application parameters such those related to timing and duration. A hybrid discrete fracture network and multi-porosity (DFN-MP) model for fluid interaction with the formation where the well is located can be produced by a processing device. The hybrid DFN-MP model can be optimized using field simulation data for the formation. The optimized hybrid DFN-MP model can be used by the processing device to produce an optimized foam-fluid-application scenario, which can be communicated to at least one well and can be utilized to stimulate the well for increased production.

Abstract: System and methods of modeling fluids in a simulation of fluid production in a multi-reservoir system with a common surface network are provided. Pressure-volume-temperature (PVT) data is determined for fluids in each of a plurality of reservoirs coupled to the common surface network. A shared equation of state (EOS) characterization representing each of the fluids across the plurality of reservoirs is generated based on the corresponding PVT data. Data representing properties of the fluids in each reservoir is calculated based on the shared EOS characterization of the fluids. When the calculated data is determined not to match the PVT data associated with the fluids in each reservoir, to the shared EOS characterization is adjusted based on a difference between the calculated data and the PVT data.

Abstract: System and methods of simulating fluid production in a multi-reservoir system with a common surface network are provided. Black oil data is matched with a common equation of state (EOS) model for each of a plurality of reservoirs. The black oil data representing fluids within each reservoir. At least one multi-dimensional black oil table representing a mix of the fluid components to be produced from each of the plurality of reservoirs via the common surface network is generated based on the EOS model that matches the one or more black oil tables for each reservoir. Properties of the fluids in the mix during a simulation of fluid production from the plurality of reservoirs are calculated based on the generated multi-dimensional black oil table for each reservoir.

Abstract: System and methods of simulating fluid production in a multi-reservoir system with a common surface network are provided. Black oil data is matched with an equation of state (EOS) model representing different fluid components of each reservoir in the multi-reservoir system. The black oil data is converted into a two-component black oil model for each reservoir, based on the EOS model. Fluid production in the multi-reservoir system is simulated for at least one simulation point in the common surface network, based in part on the two-component black oil model of each reservoir. When fluids produced at the simulation point are determined to be from different reservoirs, properties of the fluids are calculated based on weaved EOS models of the different reservoirs. Otherwise, properties of the fluids are calculated using the two-component black oil model for the reservoir from which the fluids are produced.

Abstract: A method for implementing a reservoir simulator is described. The method comprises developing training data by performing a calculation on an initial set of input data relating to reservoir conditions to obtain corresponding output data; training an artificial neural network (“ANN”) to perform the calculation using the training data; and using the trained ANN to perform the calculation on a second set of input data to obtain corresponding output data for use by the reservoir simulator in performing simulations.

Abstract: A method includes modeling a fluid flow network, the fluid flow network having a surface pipeline network connected between a plurality of well perforation nodes and a common outlet or inlet. The method also includes generating a plurality of two-phase envelopes for the modeled fluid flow network, where each two-phase envelope has at least some interpolated values and corresponds to a section of the modeled fluid flow network with a constant flow composition. The method also includes determining phase equilibrium information for the modeled fluid flow network based on the generated two-phase envelopes. The method also includes applying the determined phase equilibrium information to production or simulation related to the fluid flow network.

Abstract: System and methods of simulating fluid production in a multi-reservoir system with a common surface network are presented. An equation of state (EOS) characterization of fluids is matched with a delumped EOS model representing different components of the fluids for each reservoir within the multi-reservoir system. Fluid production in the multi-reservoir system is simulated for at least one simulation point in the common surface network, based in part on the delumped EOS model for each reservoir. If the fluids produced during the simulation at the simulation point are mixed fluids from different reservoirs, one or more interpolation tables representing the mixed fluids are generated and properties of the mixed fluids are calculated based on the generated interpolation tables. Otherwise, the properties of the fluids are calculated using the delumped EOS model corresponding to the reservoir from which the fluids are produced.

Abstract: System and methods of simulating fluid production in a multi-reservoir system with a common surface network are provided. Black oil data is matched with a common equation of state (EOS) model for each of a plurality of reservoirs. The black oil data representing fluids within each reservoir. At least one multi-dimensional black oil table representing a mix of the fluid components to be produced from each of the plurality of reservoirs via the common surface network is generated based on the EOS model that matches the one or more black oil tables for each reservoir. Properties of the fluids in the mix during a simulation of fluid production from the plurality of reservoirs are calculated based on the generated multi-dimensional black oil table for each reservoir.

Abstract: System and methods of simulating different phase states of fluids in a reservoir are presented. A phase envelope is approximated in response to a non-successful calculation of the saturation pressure. If a given temperature is above a maximum temperature of the approximate phase envelope, a phase of the grid-block is set as a single phase vapor with a composition equal to the overall composition. A saturation pressure is interpolated from the approximate phase envelope in response to a determination that the given temperature is not above the maximum temperature of the approximate phase envelope. If the interpolated saturation pressure is within an accuracy tolerance range, the phase state of the grid-block is determined based on the interpolated saturation pressure.

Abstract: System and methods of modeling fluids in a simulation of fluid production in a multi-reservoir system with a common surface network are provided. Pressure-volume-temperature (PVT) data is determined for fluids in each of a plurality of reservoirs coupled to the common surface network. A shared equation of state (EOS) characterization representing each of the fluids across the plurality of reservoirs is generated based on the corresponding PVT data. Data representing properties of the fluids in each reservoir is calculated based on the shared EOS characterization of the fluids. When the calculated data is determined not to match the PVT data associated with the fluids in each reservoir, to the shared EOS characterization is adjusted based on a difference between the calculated data and the PVT data.

Abstract: System and methods of simulating fluid production in a multi-reservoir system with a common surface network are provided. Black oil data is matched with an equation of state (EOS) model representing different fluid components of each reservoir in the multi-reservoir system. The black oil data is converted into a two-component black oil model for each reservoir, based on the EOS model. Fluid production in the multi-reservoir system is simulated for at least one simulation point in the common surface network, based in part on the two-component black oil model of each reservoir. When fluids produced at the simulation point are determined to be from different reservoirs, properties of the fluids are calculated based on weaved EOS models of the different reservoirs. Otherwise, properties of the fluids are calculated using the two-component black oil model for the reservoir from which the fluids are produced.