Premium
Evolving Connectionist Model to Monitor the Efficiency of an In Situ Combustion Process: Application to Heavy Oil Recovery
Author(s) -
Ahmadi Mohammad Ali,
Masoumi Mohammad,
Askarinezhad Reza
Publication year - 2014
Publication title -
energy technology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.91
H-Index - 44
eISSN - 2194-4296
pISSN - 2194-4288
DOI - 10.1002/ente.201402043
Subject(s) - combustion , least squares function approximation , oil field , process (computing) , process engineering , least squares support vector machine , enhanced oil recovery , artificial neural network , support vector machine , petroleum engineering , field (mathematics) , computer science , engineering , machine learning , mathematics , statistics , chemistry , organic chemistry , estimator , operating system , pure mathematics
A primary difference between conventional oil and unconventional heavy oil reservoirs is the added economic value to recovery from heavy oil reserves due to the sweep efficiency. To determine the added value, one needs to obtain the recovery factor of in situ combustion; however, this requires special experimental and laboratory combustion study and field tests. In the absence of experimental studies during the early period of field exploration, techniques that correlate such a parameter are of interest for engineers. In this work, a new method called “least‐squares support vector machine” was developed to monitor the recovery factor of the in situ combustion employment through heavy oil reservoirs. The proposed approach is applied to the experimental data from extensive works reported in the literature and the model has been implemented, developed, and tested. The predicted results from the least‐squares support vector machine model were compared to the addressed real in situ combustion data. A comparison between the generated outcomes of our model and the alternatives proves that the least‐squares support vector machine model estimates the efficiency of the in situ combustion with high degree of accuracy. The least‐squares support vector machine does not contain any conceptual errors such as over‐fitting, which can be an issue for artificial neural networks. The outcomes of this research could be coupled with commercial production software for heavy oil reservoirs to enhance production optimization and facilitate design.