Value creation is achieved only when we can make robust and high-quality decisions. The key input for decision-making is the forecast/prediction capability. Nevertheless, one can’t make good decisions without embracing uncertainties.
Forecast/Prediction competence is nowadays enabled by either models (physics based) or data driven (Machine Learning and Artificial Intelligence) or hybrid (Physics Informed Neural Networks type) methodologies.
The type of decisions ranges from asset level (single reservoir) to area development (multiple reservoirs) and further on to a higher-level portfolio/strategy optimization.
The paradigm-shit that we propose is to move gradually from the data/model/hybrid driven reservoir management towards a decision-driven reservoir management. Starting with a decision space (more than one decision, e.g., alternatives) one should be able to have access to fit-for-purpose models/data/tools tailored to the decisions in hand.
The talk will cover fundamental concepts/technology development/state-of-the-art methodologies/research/algorithms for model calibration (data assimilation), optimization (gradient based and gradient free) and a decision-making workflow which are part of our current research in UiS and their applications/implementations in real life applications.
Remus has a background in Mathematics, with an undergraduate and a MSc degree in Applied Statistics from University of Bucharest. He holds a PhD in Applied Mathematics from TU Delft, in the area of atmospheric chemistry and air pollution. Spent 7 years working as a senior researcher in the Department of Energy, The Dutch National Research Center. Since 2013, Remus is the lead scientist for the Technology Development of Reservoir Management and Production Optimization under geological uncertainties, in the Technology, Development and Innovation group. For the last 2 years he took on the Leading Advisor Assisted History Matching and Optimization position. He also has a part time professorship position in Department of Energy Resources, University of Stavanger. Remus’s main research topics are:
Remus teaches a specialized course for PhD and MSc students on Inverse Modeling, Data Assimilation and Optimization with applications in Reservoir Management and he is supervising MSc and PhD student
Lead scientist for the Technology Development of Reservoir Management and Production Optimization under geological uncertainties, in the Technology, Development and Innovation group