5 results about "Production forecasting" patented technology

Implementations described and claimed herein provide systems and methods for an innovative machine learning-driven approach, rooted in the fundamental physics of flow within fractured tight reservoirs for production forecasting of unconventional reservoirs. A first component of the method is to automatically analyze production data and generate characteristic attributes for linear flow and boundary-dominated flow. Following this, a Markov chain Monte Carlo process is utilized to integrate actual production data with flow regime analysis, resulting in probabilistic multi-segment decline models for production forecasting with uncertainty ranges and confidence estimation. Further, the method may include a two-step machine learning model to predict future planned wells. The two-step machine learning model may include a first aspect to generate predicted flow regime characteristics for one or more unconventional reservoirs and a second aspect to utilize the flow regime characteristics to generate the production forecast for the reservoirs.

This invention belongs to the field of numerical simulation technology for oil and gas reservoir development, specifically relating to a numerical simulation method for enhanced oil recovery (EOR) through carbon dioxide injection into coalbed methane. The method includes: establishing a three-dimensional geological model, a three-dimensional geomechanical model, and a fracture network model using Petrel; establishing a numerical simulation model of coalbed methane components including fractures using the Intersect simulator, and performing historical data fitting based on historical production data; importing the geological attribute field, pressure field, and saturation field at the end of the historical data fitting into the CMG-GEM simulator to conduct thermal simulation of carbon dioxide injection into coalbed methane; importing the simulated pressure and temperature fields back into Petrel, and then using the Intersect simulator to complete the dynamic prediction of thermal drive production. This invention can consider the influence of temperature on CH4 and CO2 adsorption / desorption, improving model accuracy and significantly enhancing the accuracy of production prediction.

The application discloses a kind of based on Monte Carlo simulation's bayesian random fluctuation natural gas production prediction method, it is related to natural gas production prediction technical field.The application obtains natural gas historical collection data, carries out target function construction, calculates the initial value of yield probability;According to yield probability initial value, introduce Monte Carlo error judgment mechanism, carry out mean square error estimation calculation to natural gas production by natural gas collection historical factor data;Natural gas collection historical factor data is unfolded to multiple factors, and a multiple factor judgment weight model is formed, and the multiple factor judgment weight model is predicted to natural gas production by bayesian training formula.The application can better avoid natural gas prediction data overfitting phenomenon by using Monte Carlo simulation combined with bayesian calculation, and the robustness of data is enhanced, the approximate weight distribution calculation is carried out by the whole natural gas data multiple factor calculation model, and the accuracy of natural gas prediction judgment is improved.

A method of predicting production characteristics of a hydrocarbon well using time lapse geochemistry fingerprinting and using machine learning to train a reservoir model to accurately predict production characteristics.

Implementations described and claimed herein provide systems and methods for an innovative machine learning-driven approach, rooted in the fundamental physics of flow within fractured tight reservoirs for production forecasting of unconventional reservoirs. A first component of the method is to automatically analyze production data and generate characteristic attributes for linear flow and boundary-dominated flow. Following this, a Markov chain Monte Carlo process is utilized to integrate actual production data with flow regime analysis, resulting in probabilistic multi-segment decline models for production forecasting with uncertainty ranges and confidence estimation. Further, the method may include a two-step machine learning model to predict future planned wells. The two-step machine learning model may include a first aspect to generate predicted flow regime characteristics for one or more unconventional reservoirs and a second aspect to utilize the flow regime characteristics to generate the production forecast for the reservoirs.