Logo


      You are here : Industrial Developments > Joint Industry Projects > SimuloNN


SimuloNN

Compositionnal Reservoir Simulator including a Neural Network

Objectives

SimuloNN is a plug-in for compositionnal reservoir simulators aiming to replace the flash calculations. This plug-in will be built before being linked to the simulator using the technology of neural network. It will allow capturing complex thermodynamic equilibrium through stand-alone thermodynamic packages.


Back-ground

Terra 3E has developped proxies to reproduce the results of a reservoir simulator using similar technology1.


Program

The first year of the JIP SimuloNN will be devoted to validated the methodology using the open source reservoir simulator eWoms2 with the PVT data of the SPE 5 Comparative project3.


Delivery

Results with and without SimuloNN allowing to compare the efficiency of the approach.


Start date

The JIP is planned to start in January 2014.


Useful links



For further information, documentation or to find out how to participate in this project, please click here to contact us.


1J. Bruyelle and D.R. Guérillot: Neural Networks and their Derivatives for History Matching and other Seismic, Basin and Reservoir Optimization Problems, ECMOR XIII - 13th European Conference on the Mathematics of Oil Recovery, Biarritz France, September 2012

2 http://www.opm-project.org/

3 J.E. Killough and C.A. Kossack: Fifth Comparative Solution Project: Evaluation of Miscible Flood Simulators, Ninth SPE Symposium on Reservoir Simulation, 1987



© 2009 - 2015 Terra 3E SAS