From the left: Anders Matheson, Yiteng Zhang, Rolf Johan Lorentzen, Xiaodong Luo, Tor Harald Sandve, Sergey Alyaev, Geir Evensen, Robert Klöfkorn, Anna Kvashchuk, Trine Mykkeltvedt and Yuqing Chang.
22.09.2017Eleven scientists from IRIS arranged a mini-symposium at the SIAM conference on mathematical and computational issues in the geosciences.
The conference took place September 11—14, 2017 in Erlangen, Germany and Trine S. Mykkeltvedt from the reservoir group at IRIS was one of the eleven IRIS scientists present in Erlangen. She explains that the SIAM conference is an excellent arena to learn about the latest research within advanced modeling and simulation and their application in various fields of geosciences.
Eight talks by IRIS researchers
“And not the least, it was great for IRIS to arrange a mini-symposium with three sessions where we could present some of our latest research results within reservoir management. Eight oral talks were given by IRIS researchers and in addition we co-authored more talks”, Mykkeltvedt explains.
“It is no longer merely a question how history matching seismic data could serve as an aid for reservoir management”, Mykkeltvedt claims and refers to Rolf Lorentzen’s talk Use of correlation-based localization for history matching seismic data
. “He addressed a novel approach for handling the number of data by employing correlation-based localization, in hopes to settle the issues of underestimated uncertainty and ensemble collapse when using ensemble based data assimilation techniques”, Mykkeltvedt continues.
Mykkeltvedt also emphasizes Yuqing Chang’s talk Robustness Evaluation of Ensemble-based Methods for Production Optimization
. “Yuqing, on the other hand, enriches the field of production optimization using ensemble-based methods with the use of Gaussian Mixture optimization. The results, in brief, show significant variations of net-present-value resulted from multiple factors such as stochastic perturbation of ensembles and uncertainty of geo-models”, according to Mykkeltvedt.
Yuqing Chang: Robustness Evaluation of Ensemble-based Methods for Production Optimization.
Geir Evensen revisited the basics of data assimilation and discussed what the methods actually converge to in his talk Properties of iterative ensemble smoothers for history matching
Xiaodong Luo was one of six to join the machine learning session with his talk Efficient big data assimilation through sparse representation
. He presented his work about big data problem arising from history matching using seismically imaging. This presentation demonstrated our capacity to transfer existing knowledge to the new trending field.
Trine Mykkeltvedt, Anna Kvashchuk and Robert Klöfkorn all presented their research on numerical methods related to increased oil recovery. Especially, several results on how important implicit and higher-order schemes are to simulate the flow in a reservoir as accurate and fast as possible. Their research was presented and well received.
Tor Harald Sandve presented the open source software OPM for reservoir simulation and visualization, and showed how mature the software has become. Sergey presented interesting research from his Ph.D on fractal structures in freezing brine.
Tor Harald Sandve.
“In conclusion, several researchers from the group contributed to shape the society of industrial and applied mathematics towards a more thorough interconnection between theory and practice” Trine Mykkeltvedt concludes.
All photos: Sergey Alyaev
View all conference details on SIAM's website here.