Reservoir Model Optimization
The modeling of underwater reservoirs is an important task for the Carbon Capture and Storage project, which aims at reducing CO2 emissions by storing carbon gas that has is produced by power plants in subsurface reservoirs that previously held oil, natural gas, etc.
One of the main issues of this task is to make sure that the model of the underground reservoir is correct. Only partial information, such as seismic surveys and history of well exploration data, is available to generate these models. To find out the other parameters, it is necessary to perform an inverse modeling problem where we simulate reservoirs with several parameters, and see how the results compare with the historical data (History Matching).
Of course, this is complicated by hidden information, errors in the model and in the data, etc.
To approach this problem, we use blackbox optimizers based on evolutionary algoritm. In particular, we are interested in how Multi-Objective optimization algorithms can be used to reduce the uncertainty of the model, by setting several references as different trade-off objectives.
Also, because model simulation is very time consuming (each simulation can take hours!), we must find out how to get the most out of every evaluation used by the algorithm.
Romain Chassagne, Claus Aranha, "A Pragmatic Investigation of the Objective Function for Subsurface Data Assimilation Problem", Operations Research Perspectives, Volume 7, 2020.02. DOI: 10.1016/j.orp.2020.100143
Claus Aranha, Ryoji Tanabe, Romain Chassagne and Alex Fukunaga, "Optimization of Oil Reservoir Model Using Tuned Evolutionary Algorithms and Adaptive Differential Evolution", Proceedings of the IEEE Congress on Evolutionary Computation, pp.877-884 (2015.5)
This research is a cooperation with the Institute of Petroleum Engineering from Heriot-Watt University, Edinburgh.