A Survey on the Application of Machine Learning and Metaheuristic Algorithms for Intelligent Proxy Modeling in Reservoir Simulation

•A survey of applying ML and coupled ML-metaheuristic paradigm in intelligent proxy modeling of NRS is provided.•Discussion about relevant previous works is outlined.•General methodology of developing models using ML is expounded.•Application of ML and coupled ML-metaheuristic paradigm in different...

Full description

Saved in:
Bibliographic Details
Published in:Computers & chemical engineering 2023-02, Vol.170, p.108107, Article 108107
Main Authors: Ng, Cuthbert Shang Wui, Nait Amar, Menad, Jahanbani Ghahfarokhi, Ashkan, Imsland, Lars Struen
Format: Article
Language:English
Subjects:
Data-Driven Modeling
Intelligent Proxies
Machine Learning
Metaheuristic Algorithms
Numerical Reservoir Simulation
Reservoir Engineering
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•A survey of applying ML and coupled ML-metaheuristic paradigm in intelligent proxy modeling of NRS is provided.•Discussion about relevant previous works is outlined.•General methodology of developing models using ML is expounded.•Application of ML and coupled ML-metaheuristic paradigm in different domains of reservoir simulation is discussed. Machine Learning (ML) has demonstrated its immense contribution to reservoir engineering, particularly reservoir simulation. The coupling of ML and metaheuristic algorithms illustrates huge potential for application in reservoir simulation, specifically in developing proxy models for fast reservoir simulation and optimization studies. This is conveniently termed the coupled ML-metaheuristic paradigm. Generally, proxy modeling has been extensively researched due to the expensive computational effort needed by traditional Numerical Reservoir Simulation (NRS). ML and the abovementioned coupled paradigm are effective in establishing proxy models. We conduct a survey on the employment of ML and the coupled paradigm in proxy modeling of NRS. We present the respective successful applications as reported in the literature. The benefits and limitations of these methods in intelligent proxy modeling are briefly explained. We opine that some study areas, including sampling techniques and dimensionality reduction methods, are worth investigating as part of the future research development of this technology.
ISSN:0098-1354
1873-4375
DOI:10.1016/j.compchemeng.2022.108107