Frontrunner concepts
Concept choices
Problem
Field Development Planning (FDP) is a complex and high stake decision game under uncertain conditions.
A wide range of integrated development concepts needs to be evaluated and optimised over the full range of subsurface uncertainty to understand robustness of the decisions.
Existing ways of working are often manual, slow and incomplete. In most FDP studies only a subset of concepts is evaluated against a handful of subsurface realisations.
As a result there is a high probability of a poor quality business decision and significant risk that value is left on the table.
Use our Game AI approach to consistently evaluate the full option space against the full range of subsurface uncertainty, and understand the trade offs between different value metrics.
The resulting insight can be used to make much more robust decisions that are supported by all stakeholders, or to build a data-driven case to acquire additional information.
Process
We translated the high stake development planning for a fictitious gas field into an optimisation problem, and taught an AI agent to play this game.
Through self-play, the AI learns to master the game mechanics and to develop development strategies that outperform human capability.
Using our AI as an assistant, we can now evaluate field development plans much more efficiently and effectively than before. This leads to more robust decisions that are supported by all relevant stakeholders.
Dashboard
Assume
exploration
success
Frontrunner concepts
Capital efficiency
(VIR 5%)
Net present value
(5%, mln USD)
Recovery
(mln barrels of oil equivalent)
Workflow
The AI has evaluated all options and optimised the drilling schedule for each outcome.
All concepts are evaluated & ranked. The dashboard presents the ranges of key decision metrics for the frontrunner concepts.
Use the dashboard to understand the individual concept choices and trade offs between the decision metrics.
Consider what-if scenario's such as the impact of exploration outcomes to test concept robustness.