TY - JOUR
T1 - Modelling of petroleum multiphase flow in electrical submersible pumps with shallow artificial neural networks
AU - Mohammadzaheri, Morteza
AU - Tafreshi, Reza
AU - Khan, Zurwa
AU - Ghodsi, Mojatba
AU - Franchek, Mathew
AU - Grigoriadis, Karolos
N1 - Publisher Copyright:
© 2019, © 2019 Informa UK Limited, trading as Taylor & Francis Group.
PY - 2019
Y1 - 2019
N2 - This paper first investigates existing empirical models which predict head or pressure increase of two-phase petroleum fluids in electrical submersible pumps (ESPs); then, proposes an alternative model, a shallow artificial neural network (ANN) for the same purpose. Empirical models of ESP are widely used; whereas, analytical models are still unappealing due to their reliance on over-simplified assumptions, need to excessive extent of information or lack of accuracy. The proposed shallow ANN is trained and cross-validated with the same data used in developing a number of empirical models; however, the ANN evidently outperforms those empirical models in terms of accuracy in the entire operating area. Mean of absolute prediction error of the ANN, for the experimental data not used in its training, is 69% less than the most accurate existing empirical model.
AB - This paper first investigates existing empirical models which predict head or pressure increase of two-phase petroleum fluids in electrical submersible pumps (ESPs); then, proposes an alternative model, a shallow artificial neural network (ANN) for the same purpose. Empirical models of ESP are widely used; whereas, analytical models are still unappealing due to their reliance on over-simplified assumptions, need to excessive extent of information or lack of accuracy. The proposed shallow ANN is trained and cross-validated with the same data used in developing a number of empirical models; however, the ANN evidently outperforms those empirical models in terms of accuracy in the entire operating area. Mean of absolute prediction error of the ANN, for the experimental data not used in its training, is 69% less than the most accurate existing empirical model.
KW - Electrical submersible pump
KW - empirical model
KW - multiphase petroleum fluid
KW - shallow artificial neural network
UR - http://www.scopus.com/inward/record.url?scp=85064641450&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85064641450&partnerID=8YFLogxK
U2 - 10.1080/17445302.2019.1605959
DO - 10.1080/17445302.2019.1605959
M3 - Article
AN - SCOPUS:85064641450
SN - 1744-5302
VL - 15
SP - 174
EP - 183
JO - Ships and Offshore Structures
JF - Ships and Offshore Structures
IS - 2
ER -