The Fast Reservoir Forecaster quickly predicts a field’s generative potential
A project presented by Zinyat Agharzayeva, Raquel Moussa, Jed Oukmal, Achraf Ourir and Baptiste Rondeleux
A cloud application, the Fast Reservoir Forecaster can provide reservoir engineers with reliable forecasts of the hydrocarbon resources in a reservoir. This workflow automation based on artificial intelligence allows users to save valuable time without jeopardizing quality, a decisive competitive asset.
Fast Reservoir Forecaster: no more tedious workflows
The performance of our exploration and production operations depends in large part on how well the hydrocarbon resources in our reservoirs and the related uncertainty are evaluated. It also lies in the potential for obtaining accurate production forecasts in a short amount of time, which requires a high degree of responsiveness from our reservoir engineers: this is key to business success for new-business negotiators in a context marked by unpredictable schedules and short turnaround times.
Based on a manual forecast production flow, conventional forecasting methods required up to three weeks of work. The Fast Reservoir Forecaster accelerates, improves and simplifies this tedious process.
Examining reservoirs with machine learning: a first for Exploration-Production
Although the Fast Reservoir Forecaster (FRF) is based on complex AI algorithms, it’s very easy to use.
- The application uses machine learning to generate hybrid models that are calibrated with massive historical data gathered from the field to reproduce complex physical phenomena. It provides measurability tools allowing users to test out different hypotheses and select the most representative reservoir forecasting models.
- These models combine traditional forecasting methods such as decline curves as well as advanced methods to carry out well groupings and identify the best analogies. Moreover, they make it possible to round out data that is traditionally composed only of production information, adding key business-line data generated by geoscientific overview studies.
- The application allows users to create customized workflows that meet the needs of each field, especially those with a long history and a very large number of wells.
A forecast in just three hours!
- Automating workflows simplifies a process that has always been tedious, from loading the information all the way to analyzing the results. Combined with the rapid use of massive historical data to qualify the quality of the models, it saves users time without sacrificing forecasting quality.
- With this customized web application, an initial forecast can be formulated in just three hours, leaving more time for specialized analyses with greater added value, such as uncertainty calculations.
- The tool allows engineers to adapt to a fast-paced and demanding business context, in particular for studies having to do with opportunities related to resources that have already been explored, all while ensuring the quality of the forecasting results. Users can access the Fast Reservoir Forecaster via the cloud simply by entering their login.
Developed in agile mode, the Fast Reservoir Forecaster has already been adopted by several TotalEnergies affiliates, in particular for the mature fields Waha (Libya) and Al-Shaheen (Qatar).
The project team
Exploration-Production’s essential link