Optimization of surface condensate production from natural gases using artificial intelligence

Farhan A. Al-Farhan, Luis Ayala H.

Research output: Contribution to journalArticle

8 Citations (Scopus)

Abstract

The selection of operating pressures in surface separators can have a remarkable impact on the quantity and quality of the oil produced at the stock tank. In the case of a three-stage separation process, where the operating pressures of the first and third stage (stock tank) are usually set by process considerations, the middle-stage separator pressure becomes the natural variable that lends itself to optimization. Middle-stage pressure is said to be optimum when it maximizes liquid yield in the production facility (i.e., CGR value reaches a maximum) while enhancing the quality of the produced oil condensate (i.e., API is maximized). Accurate thermodynamic and phase equilibrium calculations, albeit elaborate and computer-intensive, represent the more rigorous and reliable way of approaching this optimization problem. Nevertheless, empirical and quasi-empirical approaches are typically the norm when it comes to the selection of the middle-stage surface separation pressure in field operations. In this study, we propose the implementation of Artificial Neural Network (ANN) technology for the establishment of an expert system capable of learning the complex relationship between the input parameters and the output response of the middle-stage optimization problems via neuro-simulation. During the neuro-simulation process, parametric studies are conducted to identify the most influential variables in the thermodynamic optimization protocol. This study presents a powerful optimization tool for the selection of the optimum middle-stage separation pressure, for a variety of natural gas fluid mixtures. The developed ANN is able to predict operating conditions for optimum surface condensate recovery from typical natural gases with condensate contents between 10 < CGR < 500 STB/MMscf with stock tank API gravity ranging between 20 < API < 90, while conditions at the high-pressure separator can fluctuate between 50 and 1000 psia. Fast reliable and inexpensive results can be obtained using the proposed approach while still capturing the salient thermodynamic characteristics and without recurring to detailed-and hence expensive-phase behavior calculations.

Original languageEnglish (US)
Pages (from-to)135-147
Number of pages13
JournalJournal of Petroleum Science and Engineering
Volume53
Issue number1-2
DOIs
StatePublished - Aug 1 2006

Fingerprint

artificial intelligence
condensate
Artificial intelligence
natural gas
Natural gas
Separators
Application programming interfaces (API)
thermodynamics
artificial neural network
Thermodynamics
Neural networks
oil
expert system
phase equilibrium
Phase behavior
simulation
Phase equilibria
Expert systems
learning
Gravitation

All Science Journal Classification (ASJC) codes

  • Fuel Technology
  • Geotechnical Engineering and Engineering Geology

Cite this

@article{2c514b8f284642e8a0837dc3ab835854,
title = "Optimization of surface condensate production from natural gases using artificial intelligence",
abstract = "The selection of operating pressures in surface separators can have a remarkable impact on the quantity and quality of the oil produced at the stock tank. In the case of a three-stage separation process, where the operating pressures of the first and third stage (stock tank) are usually set by process considerations, the middle-stage separator pressure becomes the natural variable that lends itself to optimization. Middle-stage pressure is said to be optimum when it maximizes liquid yield in the production facility (i.e., CGR value reaches a maximum) while enhancing the quality of the produced oil condensate (i.e., API is maximized). Accurate thermodynamic and phase equilibrium calculations, albeit elaborate and computer-intensive, represent the more rigorous and reliable way of approaching this optimization problem. Nevertheless, empirical and quasi-empirical approaches are typically the norm when it comes to the selection of the middle-stage surface separation pressure in field operations. In this study, we propose the implementation of Artificial Neural Network (ANN) technology for the establishment of an expert system capable of learning the complex relationship between the input parameters and the output response of the middle-stage optimization problems via neuro-simulation. During the neuro-simulation process, parametric studies are conducted to identify the most influential variables in the thermodynamic optimization protocol. This study presents a powerful optimization tool for the selection of the optimum middle-stage separation pressure, for a variety of natural gas fluid mixtures. The developed ANN is able to predict operating conditions for optimum surface condensate recovery from typical natural gases with condensate contents between 10 < CGR < 500 STB/MMscf with stock tank API gravity ranging between 20 < API < 90, while conditions at the high-pressure separator can fluctuate between 50 and 1000 psia. Fast reliable and inexpensive results can be obtained using the proposed approach while still capturing the salient thermodynamic characteristics and without recurring to detailed-and hence expensive-phase behavior calculations.",
author = "Al-Farhan, {Farhan A.} and {Ayala H.}, Luis",
year = "2006",
month = "8",
day = "1",
doi = "10.1016/j.petrol.2006.05.001",
language = "English (US)",
volume = "53",
pages = "135--147",
journal = "Journal of Petroleum Science and Engineering",
issn = "0920-4105",
publisher = "Elsevier",
number = "1-2",

}

Optimization of surface condensate production from natural gases using artificial intelligence. / Al-Farhan, Farhan A.; Ayala H., Luis.

In: Journal of Petroleum Science and Engineering, Vol. 53, No. 1-2, 01.08.2006, p. 135-147.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Optimization of surface condensate production from natural gases using artificial intelligence

AU - Al-Farhan, Farhan A.

AU - Ayala H., Luis

PY - 2006/8/1

Y1 - 2006/8/1

N2 - The selection of operating pressures in surface separators can have a remarkable impact on the quantity and quality of the oil produced at the stock tank. In the case of a three-stage separation process, where the operating pressures of the first and third stage (stock tank) are usually set by process considerations, the middle-stage separator pressure becomes the natural variable that lends itself to optimization. Middle-stage pressure is said to be optimum when it maximizes liquid yield in the production facility (i.e., CGR value reaches a maximum) while enhancing the quality of the produced oil condensate (i.e., API is maximized). Accurate thermodynamic and phase equilibrium calculations, albeit elaborate and computer-intensive, represent the more rigorous and reliable way of approaching this optimization problem. Nevertheless, empirical and quasi-empirical approaches are typically the norm when it comes to the selection of the middle-stage surface separation pressure in field operations. In this study, we propose the implementation of Artificial Neural Network (ANN) technology for the establishment of an expert system capable of learning the complex relationship between the input parameters and the output response of the middle-stage optimization problems via neuro-simulation. During the neuro-simulation process, parametric studies are conducted to identify the most influential variables in the thermodynamic optimization protocol. This study presents a powerful optimization tool for the selection of the optimum middle-stage separation pressure, for a variety of natural gas fluid mixtures. The developed ANN is able to predict operating conditions for optimum surface condensate recovery from typical natural gases with condensate contents between 10 < CGR < 500 STB/MMscf with stock tank API gravity ranging between 20 < API < 90, while conditions at the high-pressure separator can fluctuate between 50 and 1000 psia. Fast reliable and inexpensive results can be obtained using the proposed approach while still capturing the salient thermodynamic characteristics and without recurring to detailed-and hence expensive-phase behavior calculations.

AB - The selection of operating pressures in surface separators can have a remarkable impact on the quantity and quality of the oil produced at the stock tank. In the case of a three-stage separation process, where the operating pressures of the first and third stage (stock tank) are usually set by process considerations, the middle-stage separator pressure becomes the natural variable that lends itself to optimization. Middle-stage pressure is said to be optimum when it maximizes liquid yield in the production facility (i.e., CGR value reaches a maximum) while enhancing the quality of the produced oil condensate (i.e., API is maximized). Accurate thermodynamic and phase equilibrium calculations, albeit elaborate and computer-intensive, represent the more rigorous and reliable way of approaching this optimization problem. Nevertheless, empirical and quasi-empirical approaches are typically the norm when it comes to the selection of the middle-stage surface separation pressure in field operations. In this study, we propose the implementation of Artificial Neural Network (ANN) technology for the establishment of an expert system capable of learning the complex relationship between the input parameters and the output response of the middle-stage optimization problems via neuro-simulation. During the neuro-simulation process, parametric studies are conducted to identify the most influential variables in the thermodynamic optimization protocol. This study presents a powerful optimization tool for the selection of the optimum middle-stage separation pressure, for a variety of natural gas fluid mixtures. The developed ANN is able to predict operating conditions for optimum surface condensate recovery from typical natural gases with condensate contents between 10 < CGR < 500 STB/MMscf with stock tank API gravity ranging between 20 < API < 90, while conditions at the high-pressure separator can fluctuate between 50 and 1000 psia. Fast reliable and inexpensive results can be obtained using the proposed approach while still capturing the salient thermodynamic characteristics and without recurring to detailed-and hence expensive-phase behavior calculations.

UR - http://www.scopus.com/inward/record.url?scp=33746625689&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33746625689&partnerID=8YFLogxK

U2 - 10.1016/j.petrol.2006.05.001

DO - 10.1016/j.petrol.2006.05.001

M3 - Article

AN - SCOPUS:33746625689

VL - 53

SP - 135

EP - 147

JO - Journal of Petroleum Science and Engineering

JF - Journal of Petroleum Science and Engineering

SN - 0920-4105

IS - 1-2

ER -