Geochemical Equilibrium determination using an artificial neural network in compositional reservoir flow simulation

Dominique Guerillot, J. Guérillot

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

The application of chemical method for hydrocarbons extraction has attracted increasing interest in the reservoir simulation community. To simulate such reactive transfer processes, compositional flows in porous media with a complex mineralogical must be coupled with the chemical equilibria in the aqueous phase and the precipitation / dissolution reactions of the minerals. The most important time consumed during reactive transport simulation is the geochemical equilibrium (about 30% to 80%). Typically, chemical equilibria are computed for each cell at each time-step by solving an equations system with the iterative Newton-Raphson method. To reduce the computation time, the number of species in solution is often reduce. However, such assumption leads to a less of accuracy of results. Instead of simplifying the geochemical model, an approach that mimic the resolution of geochemical equilibrium can be considered. The aim of the approach is to provide a substitute method to bypass the huge consuming time required to balance the chemical system. This paper focuses on the use of artificial neural networks (ANN) to replace the geochemical equilibrium package. It is widely admitted that ANN are the most efficient response surface model due to the no linear behavior of the output again the parameters. This paper presents a complete workflow for compositional reservoir simulation using an artificial neural network to determine the chemical equilibrium instead of solving equations system. This approach substantially reduces the computation time while keeping an accurate equilibrium calculation. To illustrate the proposed workflow, a case study of CO2 storage in geological formation is presented. The compositional system involves 11 aqueous species, 1 mineral component, 6 chemical equilibrium reactions and 1 mineral dissolution/precipitation reaction.

Original languageEnglish
Title of host publication16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018
PublisherEuropean Association of Geoscientists and Engineers, EAGE
ISBN (Print)9789462822603
Publication statusPublished - 1 Jan 2018
Event16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018 - Barcelona, Spain
Duration: 3 Sep 20186 Sep 2018

Other

Other16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018
CountrySpain
CityBarcelona
Period3/9/186/9/18

Fingerprint

Flow simulation
artificial neural network
Neural networks
Minerals
simulation
Dissolution
Newton-Raphson method
Hydrocarbons
Porous materials
mineral
dissolution
reactive transport
chemical method
bypass
porous medium
chemical
hydrocarbon

ASJC Scopus subject areas

  • Geotechnical Engineering and Engineering Geology
  • Geochemistry and Petrology
  • Energy Engineering and Power Technology

Cite this

Guerillot, D., & Guérillot, J. (2018). Geochemical Equilibrium determination using an artificial neural network in compositional reservoir flow simulation. In 16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018 European Association of Geoscientists and Engineers, EAGE.

Geochemical Equilibrium determination using an artificial neural network in compositional reservoir flow simulation. / Guerillot, Dominique; Guérillot, J.

16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018. European Association of Geoscientists and Engineers, EAGE, 2018.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Guerillot, D & Guérillot, J 2018, Geochemical Equilibrium determination using an artificial neural network in compositional reservoir flow simulation. in 16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018. European Association of Geoscientists and Engineers, EAGE, 16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018, Barcelona, Spain, 3/9/18.
Guerillot D, Guérillot J. Geochemical Equilibrium determination using an artificial neural network in compositional reservoir flow simulation. In 16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018. European Association of Geoscientists and Engineers, EAGE. 2018
Guerillot, Dominique ; Guérillot, J. / Geochemical Equilibrium determination using an artificial neural network in compositional reservoir flow simulation. 16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018. European Association of Geoscientists and Engineers, EAGE, 2018.
@inproceedings{82b53137b60842ed86550ca96ac2d31e,
title = "Geochemical Equilibrium determination using an artificial neural network in compositional reservoir flow simulation",
abstract = "The application of chemical method for hydrocarbons extraction has attracted increasing interest in the reservoir simulation community. To simulate such reactive transfer processes, compositional flows in porous media with a complex mineralogical must be coupled with the chemical equilibria in the aqueous phase and the precipitation / dissolution reactions of the minerals. The most important time consumed during reactive transport simulation is the geochemical equilibrium (about 30{\%} to 80{\%}). Typically, chemical equilibria are computed for each cell at each time-step by solving an equations system with the iterative Newton-Raphson method. To reduce the computation time, the number of species in solution is often reduce. However, such assumption leads to a less of accuracy of results. Instead of simplifying the geochemical model, an approach that mimic the resolution of geochemical equilibrium can be considered. The aim of the approach is to provide a substitute method to bypass the huge consuming time required to balance the chemical system. This paper focuses on the use of artificial neural networks (ANN) to replace the geochemical equilibrium package. It is widely admitted that ANN are the most efficient response surface model due to the no linear behavior of the output again the parameters. This paper presents a complete workflow for compositional reservoir simulation using an artificial neural network to determine the chemical equilibrium instead of solving equations system. This approach substantially reduces the computation time while keeping an accurate equilibrium calculation. To illustrate the proposed workflow, a case study of CO2 storage in geological formation is presented. The compositional system involves 11 aqueous species, 1 mineral component, 6 chemical equilibrium reactions and 1 mineral dissolution/precipitation reaction.",
author = "Dominique Guerillot and J. Gu{\'e}rillot",
year = "2018",
month = "1",
day = "1",
language = "English",
isbn = "9789462822603",
booktitle = "16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018",
publisher = "European Association of Geoscientists and Engineers, EAGE",

}

TY - GEN

T1 - Geochemical Equilibrium determination using an artificial neural network in compositional reservoir flow simulation

AU - Guerillot, Dominique

AU - Guérillot, J.

PY - 2018/1/1

Y1 - 2018/1/1

N2 - The application of chemical method for hydrocarbons extraction has attracted increasing interest in the reservoir simulation community. To simulate such reactive transfer processes, compositional flows in porous media with a complex mineralogical must be coupled with the chemical equilibria in the aqueous phase and the precipitation / dissolution reactions of the minerals. The most important time consumed during reactive transport simulation is the geochemical equilibrium (about 30% to 80%). Typically, chemical equilibria are computed for each cell at each time-step by solving an equations system with the iterative Newton-Raphson method. To reduce the computation time, the number of species in solution is often reduce. However, such assumption leads to a less of accuracy of results. Instead of simplifying the geochemical model, an approach that mimic the resolution of geochemical equilibrium can be considered. The aim of the approach is to provide a substitute method to bypass the huge consuming time required to balance the chemical system. This paper focuses on the use of artificial neural networks (ANN) to replace the geochemical equilibrium package. It is widely admitted that ANN are the most efficient response surface model due to the no linear behavior of the output again the parameters. This paper presents a complete workflow for compositional reservoir simulation using an artificial neural network to determine the chemical equilibrium instead of solving equations system. This approach substantially reduces the computation time while keeping an accurate equilibrium calculation. To illustrate the proposed workflow, a case study of CO2 storage in geological formation is presented. The compositional system involves 11 aqueous species, 1 mineral component, 6 chemical equilibrium reactions and 1 mineral dissolution/precipitation reaction.

AB - The application of chemical method for hydrocarbons extraction has attracted increasing interest in the reservoir simulation community. To simulate such reactive transfer processes, compositional flows in porous media with a complex mineralogical must be coupled with the chemical equilibria in the aqueous phase and the precipitation / dissolution reactions of the minerals. The most important time consumed during reactive transport simulation is the geochemical equilibrium (about 30% to 80%). Typically, chemical equilibria are computed for each cell at each time-step by solving an equations system with the iterative Newton-Raphson method. To reduce the computation time, the number of species in solution is often reduce. However, such assumption leads to a less of accuracy of results. Instead of simplifying the geochemical model, an approach that mimic the resolution of geochemical equilibrium can be considered. The aim of the approach is to provide a substitute method to bypass the huge consuming time required to balance the chemical system. This paper focuses on the use of artificial neural networks (ANN) to replace the geochemical equilibrium package. It is widely admitted that ANN are the most efficient response surface model due to the no linear behavior of the output again the parameters. This paper presents a complete workflow for compositional reservoir simulation using an artificial neural network to determine the chemical equilibrium instead of solving equations system. This approach substantially reduces the computation time while keeping an accurate equilibrium calculation. To illustrate the proposed workflow, a case study of CO2 storage in geological formation is presented. The compositional system involves 11 aqueous species, 1 mineral component, 6 chemical equilibrium reactions and 1 mineral dissolution/precipitation reaction.

UR - http://www.scopus.com/inward/record.url?scp=85054597011&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054597011&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85054597011

SN - 9789462822603

BT - 16th European Conference on the Mathematics of Oil Recovery, ECMOR 2018

PB - European Association of Geoscientists and Engineers, EAGE

ER -