Coupling numerical simulation and machine learning to model shale gas production at different time resolutions

Reservoir simulation is the most robust tool for simulating gas production from the desorption controlled and hydraulically fractured shale reservoir. Incorporation of the created massive hydraulic fractures explicitly into the simulation model is the major challenge during the model development and...

Full description

Saved in:
Bibliographic Details
Published in:Journal of natural gas science and engineering 2015-07, Vol.25 (C), p.380-392
Main Authors: Kalantari-Dahaghi, Amirmasoud, Mohaghegh, Shahab, Esmaili, Soodabeh
Format: Article
Language:English
Subjects:
History matching
Machine learning
Multilateral pad
Numerical simulation
Pattern recognition
Shale gas
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!
cited_by cdi_FETCH-LOGICAL-c375t-a3ecbe6a8a851cd6dc5bd21b10ea5512a38aa0d1acf72ec8d9fef7f4d9fc317d3
cites cdi_FETCH-LOGICAL-c375t-a3ecbe6a8a851cd6dc5bd21b10ea5512a38aa0d1acf72ec8d9fef7f4d9fc317d3
container_end_page 392
container_issue C
container_start_page 380
container_title Journal of natural gas science and engineering
container_volume 25
creator Kalantari-Dahaghi, Amirmasoud
Mohaghegh, Shahab
Esmaili, Soodabeh
description Reservoir simulation is the most robust tool for simulating gas production from the desorption controlled and hydraulically fractured shale reservoir. Incorporation of the created massive hydraulic fractures explicitly into the simulation model is the major challenge during the model development and computation phases. A pattern recognition-based proxy model is our proposed technique to overcome the aforementioned problem by modeling time successive shale gas production at the hydraulic fracture cluster level at different time resolutions (Short-, medium- and long term). Ensemble of multiple, interconnected adaptive neuro-fuzzy systems create the core for the development of the shale proxy models. In this approach, unlike reduced order models, the physics and the space-time resolution are not reduced. Instead of using pre-defined functional forms that are more frequently used to develop response surfaces, a series of machine learning algorithms that conform to the system theory are used. A history-matched Marcellus shale gas pad with six horizontal laterals and 169 clusters of hydraulic fracture is used as a base case for shale proxy model development. Additionally, several realizations are defined in order to capture the uncertainties inherent in the shale simulation model. The proxy model is validated using a blind simulation run that is not used during the model development process. The developed shale proxy model re-generates the simulation results for methane production for 169 clusters hydraulic fracture in a second with high accuracy (
doi_str_mv 10.1016/j.jngse.2015.04.018
format article
fullrecord <record><control><sourceid>elsevier_osti_</sourceid><recordid>TN_cdi_osti_scitechconnect_1251837</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><els_id>S1875510015001651</els_id><sourcerecordid>S1875510015001651</sourcerecordid><originalsourceid>FETCH-LOGICAL-c375t-a3ecbe6a8a851cd6dc5bd21b10ea5512a38aa0d1acf72ec8d9fef7f4d9fc317d3</originalsourceid><addsrcrecordid>eNp9kD1PwzAQhjOARFX6C1gs9gQ7qeswMKCKL6kSC8zW9XxuHSV2ZTtI_HsSyswtN9zzvjo9RXEjeCW42Nx1VecPiaqaC1nxdcVFe1EsRKtkKQXnV8UqpY7Po3jN1aLw2zCeeucPzI8DRYfQs-SGsYfsgmfgDRsAj84T6wmin8kc2BAMTeARemIHSOwUgxnxHMnMOGspks8su4FYpBT6cT6m6-LSQp9o9beXxefz08f2tdy9v7xtH3clNkrmEhrCPW2ghVYKNBuDcm9qsRecQEpRQ9MCcCMAraoJW3NvySq7njY2QplmWdyee0PKTid0mfCIwXvCrEUtRduoCWrOEMaQUiSrT9ENEL-14HrWqTv9q1PPOjVf60nnlHo4p2j6_8tRnOvJIxkX53YT3L_5H0oRhPM</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>Coupling numerical simulation and machine learning to model shale gas production at different time resolutions</title><source>ScienceDirect Freedom Collection 2022-2024</source><creator>Kalantari-Dahaghi, Amirmasoud ; Mohaghegh, Shahab ; Esmaili, Soodabeh</creator><creatorcontrib>Kalantari-Dahaghi, Amirmasoud ; Mohaghegh, Shahab ; Esmaili, Soodabeh</creatorcontrib><description>Reservoir simulation is the most robust tool for simulating gas production from the desorption controlled and hydraulically fractured shale reservoir. Incorporation of the created massive hydraulic fractures explicitly into the simulation model is the major challenge during the model development and computation phases. A pattern recognition-based proxy model is our proposed technique to overcome the aforementioned problem by modeling time successive shale gas production at the hydraulic fracture cluster level at different time resolutions (Short-, medium- and long term). Ensemble of multiple, interconnected adaptive neuro-fuzzy systems create the core for the development of the shale proxy models. In this approach, unlike reduced order models, the physics and the space-time resolution are not reduced. Instead of using pre-defined functional forms that are more frequently used to develop response surfaces, a series of machine learning algorithms that conform to the system theory are used. A history-matched Marcellus shale gas pad with six horizontal laterals and 169 clusters of hydraulic fracture is used as a base case for shale proxy model development. Additionally, several realizations are defined in order to capture the uncertainties inherent in the shale simulation model. The proxy model is validated using a blind simulation run that is not used during the model development process. The developed shale proxy model re-generates the simulation results for methane production for 169 clusters hydraulic fracture in a second with high accuracy (&lt;15% error), thus making a comprehensive analysis of production from shale a practical and feasible option. Moreover, it can be used as an assisted history-matching tool, since it has the capability to scrutinize the solution space and produce the model response to the uncertainty changes very fast. •Ensemble of multiple, interconnected adaptive neuro-fuzzy systems created the core for the shale proxy models development.•Unlike reduced order models, the physics and the space-time resolution are not reduced.•Instead of using pre-defined functional forms that are frequently used to develop response surfaces, a series of machine learning algorithms were used.•A history-matched Marcellus shale gas pad with six horizontal laterals is used as a basis for shale proxy model development.•The developed shale proxy model re-generates the simulation results at the hydraulic fracture cluster level with good accuracy.</description><identifier>ISSN: 1875-5100</identifier><identifier>DOI: 10.1016/j.jngse.2015.04.018</identifier><language>eng</language><publisher>Netherlands: Elsevier B.V</publisher><subject>History matching ; Machine learning ; Multilateral pad ; Numerical simulation ; Pattern recognition ; Shale gas</subject><ispartof>Journal of natural gas science and engineering, 2015-07, Vol.25 (C), p.380-392</ispartof><rights>2015 Elsevier B.V.</rights><lds50>peer_reviewed</lds50><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed><citedby>FETCH-LOGICAL-c375t-a3ecbe6a8a851cd6dc5bd21b10ea5512a38aa0d1acf72ec8d9fef7f4d9fc317d3</citedby><cites>FETCH-LOGICAL-c375t-a3ecbe6a8a851cd6dc5bd21b10ea5512a38aa0d1acf72ec8d9fef7f4d9fc317d3</cites></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>230,314,780,784,885,27924,27925</link.rule.ids><backlink>$$Uhttps://www.osti.gov/biblio/1251837$$D View this record in Osti.gov$$Hfree_for_read</backlink></links><search><creatorcontrib>Kalantari-Dahaghi, Amirmasoud</creatorcontrib><creatorcontrib>Mohaghegh, Shahab</creatorcontrib><creatorcontrib>Esmaili, Soodabeh</creatorcontrib><title>Coupling numerical simulation and machine learning to model shale gas production at different time resolutions</title><title>Journal of natural gas science and engineering</title><description>Reservoir simulation is the most robust tool for simulating gas production from the desorption controlled and hydraulically fractured shale reservoir. Incorporation of the created massive hydraulic fractures explicitly into the simulation model is the major challenge during the model development and computation phases. A pattern recognition-based proxy model is our proposed technique to overcome the aforementioned problem by modeling time successive shale gas production at the hydraulic fracture cluster level at different time resolutions (Short-, medium- and long term). Ensemble of multiple, interconnected adaptive neuro-fuzzy systems create the core for the development of the shale proxy models. In this approach, unlike reduced order models, the physics and the space-time resolution are not reduced. Instead of using pre-defined functional forms that are more frequently used to develop response surfaces, a series of machine learning algorithms that conform to the system theory are used. A history-matched Marcellus shale gas pad with six horizontal laterals and 169 clusters of hydraulic fracture is used as a base case for shale proxy model development. Additionally, several realizations are defined in order to capture the uncertainties inherent in the shale simulation model. The proxy model is validated using a blind simulation run that is not used during the model development process. The developed shale proxy model re-generates the simulation results for methane production for 169 clusters hydraulic fracture in a second with high accuracy (&lt;15% error), thus making a comprehensive analysis of production from shale a practical and feasible option. Moreover, it can be used as an assisted history-matching tool, since it has the capability to scrutinize the solution space and produce the model response to the uncertainty changes very fast. •Ensemble of multiple, interconnected adaptive neuro-fuzzy systems created the core for the shale proxy models development.•Unlike reduced order models, the physics and the space-time resolution are not reduced.•Instead of using pre-defined functional forms that are frequently used to develop response surfaces, a series of machine learning algorithms were used.•A history-matched Marcellus shale gas pad with six horizontal laterals is used as a basis for shale proxy model development.•The developed shale proxy model re-generates the simulation results at the hydraulic fracture cluster level with good accuracy.</description><subject>History matching</subject><subject>Machine learning</subject><subject>Multilateral pad</subject><subject>Numerical simulation</subject><subject>Pattern recognition</subject><subject>Shale gas</subject><issn>1875-5100</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2015</creationdate><recordtype>article</recordtype><recordid>eNp9kD1PwzAQhjOARFX6C1gs9gQ7qeswMKCKL6kSC8zW9XxuHSV2ZTtI_HsSyswtN9zzvjo9RXEjeCW42Nx1VecPiaqaC1nxdcVFe1EsRKtkKQXnV8UqpY7Po3jN1aLw2zCeeucPzI8DRYfQs-SGsYfsgmfgDRsAj84T6wmin8kc2BAMTeARemIHSOwUgxnxHMnMOGspks8su4FYpBT6cT6m6-LSQp9o9beXxefz08f2tdy9v7xtH3clNkrmEhrCPW2ghVYKNBuDcm9qsRecQEpRQ9MCcCMAraoJW3NvySq7njY2QplmWdyee0PKTid0mfCIwXvCrEUtRduoCWrOEMaQUiSrT9ENEL-14HrWqTv9q1PPOjVf60nnlHo4p2j6_8tRnOvJIxkX53YT3L_5H0oRhPM</recordid><startdate>201507</startdate><enddate>201507</enddate><creator>Kalantari-Dahaghi, Amirmasoud</creator><creator>Mohaghegh, Shahab</creator><creator>Esmaili, Soodabeh</creator><general>Elsevier B.V</general><general>Elsevier</general><scope>AAYXX</scope><scope>CITATION</scope><scope>OTOTI</scope></search><sort><creationdate>201507</creationdate><title>Coupling numerical simulation and machine learning to model shale gas production at different time resolutions</title><author>Kalantari-Dahaghi, Amirmasoud ; Mohaghegh, Shahab ; Esmaili, Soodabeh</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-c375t-a3ecbe6a8a851cd6dc5bd21b10ea5512a38aa0d1acf72ec8d9fef7f4d9fc317d3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2015</creationdate><topic>History matching</topic><topic>Machine learning</topic><topic>Multilateral pad</topic><topic>Numerical simulation</topic><topic>Pattern recognition</topic><topic>Shale gas</topic><toplevel>peer_reviewed</toplevel><toplevel>online_resources</toplevel><creatorcontrib>Kalantari-Dahaghi, Amirmasoud</creatorcontrib><creatorcontrib>Mohaghegh, Shahab</creatorcontrib><creatorcontrib>Esmaili, Soodabeh</creatorcontrib><collection>CrossRef</collection><collection>OSTI.GOV</collection><jtitle>Journal of natural gas science and engineering</jtitle></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Kalantari-Dahaghi, Amirmasoud</au><au>Mohaghegh, Shahab</au><au>Esmaili, Soodabeh</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>Coupling numerical simulation and machine learning to model shale gas production at different time resolutions</atitle><jtitle>Journal of natural gas science and engineering</jtitle><date>2015-07</date><risdate>2015</risdate><volume>25</volume><issue>C</issue><spage>380</spage><epage>392</epage><pages>380-392</pages><issn>1875-5100</issn><abstract>Reservoir simulation is the most robust tool for simulating gas production from the desorption controlled and hydraulically fractured shale reservoir. Incorporation of the created massive hydraulic fractures explicitly into the simulation model is the major challenge during the model development and computation phases. A pattern recognition-based proxy model is our proposed technique to overcome the aforementioned problem by modeling time successive shale gas production at the hydraulic fracture cluster level at different time resolutions (Short-, medium- and long term). Ensemble of multiple, interconnected adaptive neuro-fuzzy systems create the core for the development of the shale proxy models. In this approach, unlike reduced order models, the physics and the space-time resolution are not reduced. Instead of using pre-defined functional forms that are more frequently used to develop response surfaces, a series of machine learning algorithms that conform to the system theory are used. A history-matched Marcellus shale gas pad with six horizontal laterals and 169 clusters of hydraulic fracture is used as a base case for shale proxy model development. Additionally, several realizations are defined in order to capture the uncertainties inherent in the shale simulation model. The proxy model is validated using a blind simulation run that is not used during the model development process. The developed shale proxy model re-generates the simulation results for methane production for 169 clusters hydraulic fracture in a second with high accuracy (&lt;15% error), thus making a comprehensive analysis of production from shale a practical and feasible option. Moreover, it can be used as an assisted history-matching tool, since it has the capability to scrutinize the solution space and produce the model response to the uncertainty changes very fast. •Ensemble of multiple, interconnected adaptive neuro-fuzzy systems created the core for the shale proxy models development.•Unlike reduced order models, the physics and the space-time resolution are not reduced.•Instead of using pre-defined functional forms that are frequently used to develop response surfaces, a series of machine learning algorithms were used.•A history-matched Marcellus shale gas pad with six horizontal laterals is used as a basis for shale proxy model development.•The developed shale proxy model re-generates the simulation results at the hydraulic fracture cluster level with good accuracy.</abstract><cop>Netherlands</cop><pub>Elsevier B.V</pub><doi>10.1016/j.jngse.2015.04.018</doi><tpages>13</tpages><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier ISSN: 1875-5100
ispartof Journal of natural gas science and engineering, 2015-07, Vol.25 (C), p.380-392
issn 1875-5100
language eng
recordid cdi_osti_scitechconnect_1251837
source ScienceDirect Freedom Collection 2022-2024
subjects History matching
Machine learning
Multilateral pad
Numerical simulation
Pattern recognition
Shale gas
title Coupling numerical simulation and machine learning to model shale gas production at different time resolutions
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2024-12-29T01%3A55%3A58IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-elsevier_osti_&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=Coupling%20numerical%20simulation%20and%20machine%20learning%20to%20model%20shale%20gas%20production%20at%20different%20time%20resolutions&rft.jtitle=Journal%20of%20natural%20gas%20science%20and%20engineering&rft.au=Kalantari-Dahaghi,%20Amirmasoud&rft.date=2015-07&rft.volume=25&rft.issue=C&rft.spage=380&rft.epage=392&rft.pages=380-392&rft.issn=1875-5100&rft_id=info:doi/10.1016/j.jngse.2015.04.018&rft_dat=%3Celsevier_osti_%3ES1875510015001651%3C/elsevier_osti_%3E%3Cgrp_id%3Ecdi_FETCH-LOGICAL-c375t-a3ecbe6a8a851cd6dc5bd21b10ea5512a38aa0d1acf72ec8d9fef7f4d9fc317d3%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true