July 30, 2024
The concept of integrated reservoir modeling is not new. However, integrated modeling goes beyond better management of interfaces between disciplines or using a single platform. It delivers a paradigm shift by integrating available data into a consistent workflow, creating models that reasonably represent the subsurface based on all available data. Because subsurface sampling often remains sparse and subject to many uncertainties, truly integrated reservoir modeling benefits from ensemble-based methods. These methods form the basis for ResX and IRMA software.
ResX software helps subsurface teams embrace uncertainties in reservoir modeling and history matching. Built from the ground up on ensemble-based methods, ResX utilizes insights from a full range of equally probable, geologically consistent reservoir models that are history-matched and can be used for production forecasts. By continuously conditioning static and dynamic data, ResX dramatically increases your team’s efficiency (Figure 1), freeing up time to explore better reservoir management options or perform more analysis to learn about your reservoir.
A successful application of ResX requires an initial ensemble of models that captures key uncertainties and is equally plausible given both measured static and dynamic data.
Create an initial ensemble
Ensemble-based modeling requires investing time to create a robust initial ensemble (Figure 2). The Create Initial Ensemble process in ResX, along with automated workflows, greatly speeds up generating a prior collection of models for history matching in an ensemble study.
The Adaptive Pluri-Gaussian (APG) facies modeling method (Figure 3) within the Create Initial Ensemble process integrates both static and dynamic data while ensuring geological consistency. During an ensemble study, each facies realization is conditioned to the dynamic data by updating the inputs to the facies modeling algorithm, ensuring consistent and geologically reasonable history-matched models.
Ensemble-based study
ResX works equally well on small fields with limited data measurements and giant fields with decades of production data. ResX scales by using a geostatistical approach to modeling, combining ensemble Kalman principles with an iterative smoother.
Before running a history matching study and production forecast, two important steps must be completed: 1) Set up the objective function and 2) Specify the model uncertainties and localization setup. From greenfields with a limited number of wells to brownfields with decades of production from hundreds of wells, the objective function in ResX supports well production data, RFT and PLT logs, DST data, and 4D seismic data.
An important step in an ensemble study is defining the localization setup, which specifies areas that can be updated when conditioning to dynamic data. This can be done manually by specifying a radius or a custom localization property, or the software can apply auto localization.
With the necessary parameters set up, the simulations can run using a dedicated simulation manager that submits and monitors them. Studies that have been stopped can be resumed, and new studies can be created from any previous iteration.
Ensemble analysis
The main goal of an ensemble-based study is to gain new insights into the reservoir. This requires evaluating the statistical attributes of properties and scalars of the models in the ensemble and comparing these to the statistics of the initial ensemble (Figure 4a). ResX provides tools to analyze and understand the initial and the resulting conditioned ensemble (Figure 4b).
These ensembles can then be automatically ingested by IRMA, an innovative, uncertainty-centric platform that manages ensemble data to generate aggregated statistics and provide specific ensemble-oriented analytics. These analytics inform reservoir management and optimization using model ensembles.
The aggregated grid properties process in ResX evaluates the statistical properties of the initial ensemble for the model properties and scalars. This helps ensure that the values of the prior distribution align with the model assumptions. The ensemble analysis process runs QC on the objective function and checks that key trends are captured.
Extracting insights from a conditioned ensemble begins with understanding what has been updated and where these updates occurred in the models. This provides valuable geological and engineering insights that can help drive improvements to modeling workflows and, more importantly, field development.
Reservoir forecasting
The main driver for building an ensemble of models is to support important field development decisions throughout an asset’s lifetime. Generating forecasts with ResX provides a better estimation of the spread of key reservoir performance indicators, helping companies make better decisions with an understanding of the risks. ResX offers tools to run these forecasts in conjunction with IRMA (Figure 5).
a) Bottom hole pressures
b) Water-cut using ResX and IRMA
The integration of ResX and IRMA delivers speed, control, and power to analyze available data. It allows immediate assessment of how modeling choices and data interpretation affect the resulting ensembles of models, enabling teams to quickly iterate, continuously learn, and improve (Figure 6).
The uncertainty-centric approach based on an end-to-end integrated and automated workflow effectively connects the input data to the ensemble, providing analytics for insights and decision support.