ISI Logo

Top-Down, Intelligent Reservoir Models™

Building coupled Reservoir-Wellbore Models Using Artificial Intelligence and Machine Learning.

Top-Down Modeling (TDM) is a realistic, tested, and validated alternative to traditional Numerical Reservoir Simulation and Modeling. TDM is a technology that combines reservoir engineering, reservoir modeling, and reservoir management with state of the art Artificial Intelligence and Machine Learning in order to build a thorough and comprehensive, full field, coupled, reservoir-wellbore model.

During the development of the Top-Down Model for a given mature field all the wells in the field are history matched. The history matching is performed simultaneously for all dynamic variable such as Oil rate, Gas, and Water production, as wellas Resevoir Pressure, and Water Saturation. Follwoing is an example of such a history match for one of the wells in a mature field in the Middle East.

Automated History Matching

The history matching process included in the Top-Down Modeling workflow is a completely automated process. This is one of the main reasons behind the fact that complete development and validation of a Top-Down Model takes a fraction of the time and resources required for numerical reservoir simulation. Another unique characteristic of the Top-Down Modeling history matching process that is not commonly practiced during the history matching of numerical reservoir simulation model is "Blind Validation in Time", and "Blind Validation in Space".

Blind Validation in Time: During the "Blind Validation in Time", several months of data from the tail-end of the production process is removed from the dataset that is used for the Top-Down Model development and history matching process. Once the history matching process is completed, the developed Top-Down Model is deployed in the forecast mode in order to predict the (oil, gas, and water) production values for the months that were left out of the process. Comparing the Top-Down Modeling forecast (predictions) with the actual left-out production values are then used in order to validate the TDM's forecast accuracy. Following figure edmonstrates examples of blind history matching in time for two offshore wells from a mature field in the Middle East.

Blind Validation in Space: During the "Blind Validation in Space", complete data from several wells are removed from the dataset before the development and history matching process of the Top-Down Model. Once the history matching process is completed, the developed Top-Down Model is deployed in the forecast mode in order to predict the (oil, gas, and water) production values for the wells that have been removed prior to the model development. Comparing the actual production values with the Top-Down Modeling forecast (predictions) of the production values of the wells that were removed prior to the model development validates the accuracy of the TDM's forecast in space. Following figure edmonstrates examples of blind history matching in space for two offshore wells from a mature field in the Middle East.

Real World Validation of the Top-Down Models

Validation of Top-Down Model's predictions were performed five years after the development for a prolific and highly complex mature field in the Middle East. The amazing performance of the Top-Down Model for accurately predicting the reservoir pressure and water saturation in that filed for infill drilling purposes is a sound proof of the capability of this AI and Machine Learning technology for subsurface modeling.