Real-time Estimation of Elastic Properties of Formation Rocks Based on Drilling Data by Using an Artificial Neural Network

E. Jamshidi, R. Arabjamaloei, A. Hashemi, M. A. Ekramzadeh, Mahmood Amani

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Understanding the mechanical properties of formations has its importance in the drilling, production, and management phases of reservoirs. In petroleum engineering, measurements through wire-line logs, which run in the boreholes after the drilling phase, are of the most common methods to estimate mechanical properties of different layers. Elastic rock mechanic measurements from logs are dynamic values and need to be upscaled and calibrated to fit the corresponding pseudo-static measurements that are obtained from cores in the laboratory. As a practical approach to having a continuous profile of these static parameters and surmount the arduousness that confronts by using core samples and laboratory tests, many researchers tried to deploy predictive methods and empirical correlations. However, it will be a great advantage to have a real-time estimation of these static parameters while drilling based on bit and formation rock interaction. Artificial neural networks are powerful tools for estimation of complex functions subjected to availability of large enough data sets that present samples of actual behavior of the function. In this study, artificial neural networks were implemented to estimate in situ rock mechanical properties, including Unconfined Compressive Strength, Young's Modulus of Elasticity, and the ratio of these parameters known as Modulus Ratio, by using operational drilling parameters as inputs. Required data were gathered from drilling reports, logging operations, and core samples from nine wells placed in Ahwaz Oilfield located in south-west Iran. The trained networks showed satisfactory low errors in the testing process and sketched the capability of artificial neural networks in estimation of complex functions. The accuracy of the presented model was then compared with the results of calibrated regional correlations and modified Warren's equation. These correlations are used in the candidate wells to estimate Unconfined Compressive Strength and Young's Modulus of Elasticity. It was observed that the new artificial neural network approach is a competent and accurate method for real-time calculation of static elastic properties of formation rocks. The results of this work could be used for drilling optimization, reducing stability problems, and lithologic boundary detection while drilling.

Original languageEnglish
Pages (from-to)337-351
Number of pages15
JournalEnergy Sources, Part A: Recovery, Utilization and Environmental Effects
Volume35
Issue number4
DOIs
Publication statusPublished - 15 Feb 2013
Externally publishedYes

Fingerprint

Drilling
Rocks
Neural networks
Elastic moduli
Core samples
Mechanical properties
Compressive strength
Petroleum engineering
Well logging
Rock mechanics
Petroleum reservoirs
Boreholes
Availability
Wire
Testing

Keywords

  • artificial neural networks
  • drilling data
  • dynamic elastic properties
  • elastic properties of rock
  • ROP models
  • static elastic properties

ASJC Scopus subject areas

  • Renewable Energy, Sustainability and the Environment
  • Nuclear Energy and Engineering
  • Fuel Technology
  • Energy Engineering and Power Technology

Cite this

Real-time Estimation of Elastic Properties of Formation Rocks Based on Drilling Data by Using an Artificial Neural Network. / Jamshidi, E.; Arabjamaloei, R.; Hashemi, A.; Ekramzadeh, M. A.; Amani, Mahmood.

In: Energy Sources, Part A: Recovery, Utilization and Environmental Effects, Vol. 35, No. 4, 15.02.2013, p. 337-351.

Research output: Contribution to journalArticle

@article{444d72be9b7f498da127ffd2d75aecfa,
title = "Real-time Estimation of Elastic Properties of Formation Rocks Based on Drilling Data by Using an Artificial Neural Network",
abstract = "Understanding the mechanical properties of formations has its importance in the drilling, production, and management phases of reservoirs. In petroleum engineering, measurements through wire-line logs, which run in the boreholes after the drilling phase, are of the most common methods to estimate mechanical properties of different layers. Elastic rock mechanic measurements from logs are dynamic values and need to be upscaled and calibrated to fit the corresponding pseudo-static measurements that are obtained from cores in the laboratory. As a practical approach to having a continuous profile of these static parameters and surmount the arduousness that confronts by using core samples and laboratory tests, many researchers tried to deploy predictive methods and empirical correlations. However, it will be a great advantage to have a real-time estimation of these static parameters while drilling based on bit and formation rock interaction. Artificial neural networks are powerful tools for estimation of complex functions subjected to availability of large enough data sets that present samples of actual behavior of the function. In this study, artificial neural networks were implemented to estimate in situ rock mechanical properties, including Unconfined Compressive Strength, Young's Modulus of Elasticity, and the ratio of these parameters known as Modulus Ratio, by using operational drilling parameters as inputs. Required data were gathered from drilling reports, logging operations, and core samples from nine wells placed in Ahwaz Oilfield located in south-west Iran. The trained networks showed satisfactory low errors in the testing process and sketched the capability of artificial neural networks in estimation of complex functions. The accuracy of the presented model was then compared with the results of calibrated regional correlations and modified Warren's equation. These correlations are used in the candidate wells to estimate Unconfined Compressive Strength and Young's Modulus of Elasticity. It was observed that the new artificial neural network approach is a competent and accurate method for real-time calculation of static elastic properties of formation rocks. The results of this work could be used for drilling optimization, reducing stability problems, and lithologic boundary detection while drilling.",
keywords = "artificial neural networks, drilling data, dynamic elastic properties, elastic properties of rock, ROP models, static elastic properties",
author = "E. Jamshidi and R. Arabjamaloei and A. Hashemi and Ekramzadeh, {M. A.} and Mahmood Amani",
year = "2013",
month = "2",
day = "15",
doi = "10.1080/15567036.2010.495971",
language = "English",
volume = "35",
pages = "337--351",
journal = "Energy Sources, Part A: Recovery, Utilization and Environmental Effects",
issn = "1556-7036",
publisher = "Taylor and Francis Ltd.",
number = "4",

}

TY - JOUR

T1 - Real-time Estimation of Elastic Properties of Formation Rocks Based on Drilling Data by Using an Artificial Neural Network

AU - Jamshidi, E.

AU - Arabjamaloei, R.

AU - Hashemi, A.

AU - Ekramzadeh, M. A.

AU - Amani, Mahmood

PY - 2013/2/15

Y1 - 2013/2/15

N2 - Understanding the mechanical properties of formations has its importance in the drilling, production, and management phases of reservoirs. In petroleum engineering, measurements through wire-line logs, which run in the boreholes after the drilling phase, are of the most common methods to estimate mechanical properties of different layers. Elastic rock mechanic measurements from logs are dynamic values and need to be upscaled and calibrated to fit the corresponding pseudo-static measurements that are obtained from cores in the laboratory. As a practical approach to having a continuous profile of these static parameters and surmount the arduousness that confronts by using core samples and laboratory tests, many researchers tried to deploy predictive methods and empirical correlations. However, it will be a great advantage to have a real-time estimation of these static parameters while drilling based on bit and formation rock interaction. Artificial neural networks are powerful tools for estimation of complex functions subjected to availability of large enough data sets that present samples of actual behavior of the function. In this study, artificial neural networks were implemented to estimate in situ rock mechanical properties, including Unconfined Compressive Strength, Young's Modulus of Elasticity, and the ratio of these parameters known as Modulus Ratio, by using operational drilling parameters as inputs. Required data were gathered from drilling reports, logging operations, and core samples from nine wells placed in Ahwaz Oilfield located in south-west Iran. The trained networks showed satisfactory low errors in the testing process and sketched the capability of artificial neural networks in estimation of complex functions. The accuracy of the presented model was then compared with the results of calibrated regional correlations and modified Warren's equation. These correlations are used in the candidate wells to estimate Unconfined Compressive Strength and Young's Modulus of Elasticity. It was observed that the new artificial neural network approach is a competent and accurate method for real-time calculation of static elastic properties of formation rocks. The results of this work could be used for drilling optimization, reducing stability problems, and lithologic boundary detection while drilling.

AB - Understanding the mechanical properties of formations has its importance in the drilling, production, and management phases of reservoirs. In petroleum engineering, measurements through wire-line logs, which run in the boreholes after the drilling phase, are of the most common methods to estimate mechanical properties of different layers. Elastic rock mechanic measurements from logs are dynamic values and need to be upscaled and calibrated to fit the corresponding pseudo-static measurements that are obtained from cores in the laboratory. As a practical approach to having a continuous profile of these static parameters and surmount the arduousness that confronts by using core samples and laboratory tests, many researchers tried to deploy predictive methods and empirical correlations. However, it will be a great advantage to have a real-time estimation of these static parameters while drilling based on bit and formation rock interaction. Artificial neural networks are powerful tools for estimation of complex functions subjected to availability of large enough data sets that present samples of actual behavior of the function. In this study, artificial neural networks were implemented to estimate in situ rock mechanical properties, including Unconfined Compressive Strength, Young's Modulus of Elasticity, and the ratio of these parameters known as Modulus Ratio, by using operational drilling parameters as inputs. Required data were gathered from drilling reports, logging operations, and core samples from nine wells placed in Ahwaz Oilfield located in south-west Iran. The trained networks showed satisfactory low errors in the testing process and sketched the capability of artificial neural networks in estimation of complex functions. The accuracy of the presented model was then compared with the results of calibrated regional correlations and modified Warren's equation. These correlations are used in the candidate wells to estimate Unconfined Compressive Strength and Young's Modulus of Elasticity. It was observed that the new artificial neural network approach is a competent and accurate method for real-time calculation of static elastic properties of formation rocks. The results of this work could be used for drilling optimization, reducing stability problems, and lithologic boundary detection while drilling.

KW - artificial neural networks

KW - drilling data

KW - dynamic elastic properties

KW - elastic properties of rock

KW - ROP models

KW - static elastic properties

UR - http://www.scopus.com/inward/record.url?scp=84872437047&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84872437047&partnerID=8YFLogxK

U2 - 10.1080/15567036.2010.495971

DO - 10.1080/15567036.2010.495971

M3 - Article

VL - 35

SP - 337

EP - 351

JO - Energy Sources, Part A: Recovery, Utilization and Environmental Effects

JF - Energy Sources, Part A: Recovery, Utilization and Environmental Effects

SN - 1556-7036

IS - 4

ER -