Everest™: An Open-Source Platform for Decision Making Under Uncertainties

The development of subsurface reservoirs involves complex decision-making process. The success of field development projects depends on the ability of making good techno-economic decisions based on currently available information about the subsurface. Multi-million investment decisions are made in the presence of uncertainties. To assist decision makers in this task, Equinor and TNO have been jointly developing the EVEREST™ framework - a decision support platform employing a flexible and computationally efficient model-based methodology for optimization under uncertainties. Uncertainties are captured using an ensemble of model realizations. The optimization can then search for optimal solutions improving the objective function evaluated over the ensemble of model realizations. The procedure delivers a single optimized solution, simultaneously optimal across all the different model realizations composing the ensemble. The practical applicability to real-life optimization challenges poses a need for computational efficiency and for supporting a vast amount of problems, e.g., different decision types, multiple objectives, nonlinear constraints, etc. The state-of-the-art stochastic gradient method at the core of EVEREST renders robust optimization over a multitude of model scenarios computationally viable. Uncertainty is not an after-thought and is accounted for throughout the entire optimization procedure. Discrete specific methods and population-based optimizers extend their applicability. The optimization engine is modular and can be extended to include more algorithms. EVEREST includes functionalities for multi-objective optimization (e.g., net present value vs. CO2 emissions) using weighted objectives, different continuous or discrete decision variables, as drilling sequence, drainage strategies and well trajectory with the application of effective mathematical parameterizations. It considers realistic operational restrictions (surface facility capacities, drilling constraints) with the usage of input or non-linear output constraints. EVEREST facilitates the work of engineers by providing a structure for the optimization problem to be framed, specifying the decision variables, objective function, uncertainties and optionally constraints. It is integrated with industry standard reservoir simulators and allows to connect to external software. Over the past decade, EVEREST has been applied to numerous real-life applications both in field development and reservoir management for green and brown fields. As a mature technology, it has been qualified as part of the software toolbox for decision support within Equinor. Its flexibility has also allowed it to be applied beyond the oil and gas domain, e.g., to optimize the layout of offshore wind farms, geothermal field development strategies, and the design of solar parks and materials and structures. After years of research and development the EVEREST tool has recently been released as open-source, providing access to cutting-edge optimization technology to a broader range of end-users and enabling open collaboration between developers and users, vendors and academic institutions, partners and competitors, within and outside the oil and gas industry. Copyright 2025, Society of Petroleum Engineers.