skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Statistical Analysis of CO 2 Exposed Wells to Predict Long Term Leakage through the Development of an Integrated Neural-Genetic Algorithm

Abstract

The objective of this project is to develop a computerized statistical model with the Integrated Neural-Genetic Algorithm (INGA) for predicting the probability of long-term leak of wells in CO 2 sequestration operations. This object has been accomplished by conducting research in three phases: 1) data mining of CO 2-explosed wells, 2) INGA computer model development, and 3) evaluation of the predictive performance of the computer model with data from field tests. Data mining was conducted for 510 wells in two CO 2 sequestration projects in the Texas Gulf Coast region. They are the Hasting West field and Oyster Bayou field in the Southern Texas. Missing wellbore integrity data were estimated using an analytical and Finite Element Method (FEM) model. The INGA was first tested for performances of convergence and computing efficiency with the obtained data set of high dimension. It was concluded that the INGA can handle the gathered data set with good accuracy and reasonable computing time after a reduction of dimension with a grouping mechanism. A computerized statistical model with the INGA was then developed based on data pre-processing and grouping. Comprehensive training and testing of the model were carried out to ensure that the model is accuratemore » and efficient enough for predicting the probability of long-term leak of wells in CO 2 sequestration operations. The Cranfield in the southern Mississippi was select as the test site. Observation wells CFU31F2 and CFU31F3 were used for pressure-testing, formation-logging, and cement-sampling. Tools run in the wells include Isolation Scanner, Slim Cement Mapping Tool (SCMT), Cased Hole Formation Dynamics Tester (CHDT), and Mechanical Sidewall Coring Tool (MSCT). Analyses of the obtained data indicate no leak of CO 2 cross the cap zone while it is evident that the well cement sheath was invaded by the CO 2 from the storage zone. This observation is consistent with the result predicted by the INGA model which indicates the well has a CO 2 leak-safe probability of 72%. This comparison implies that the developed INGA model is valid for future use in predicting well leak probability.« less

Authors:
 [1];  [2];  [3]
  1. Univ. of Louisiana, Lafayette, LA (United States)
  2. Battelle, Columbus, OH (United States)
  3. Missouri Univ. of Science and Technology, Rolla, MO (United States)
Publication Date:
Research Org.:
Univ. of Louisiana, Lafayette, LA (United States)
Sponsoring Org.:
USDOE Office of Fossil Energy (FE), Oil and Natural Gas (FE-30)
OSTI Identifier:
1373948
Report Number(s):
FinalTechnicalReport-Revised30July2017
DOE Contract Number:  
FE0009284
Resource Type:
Technical Report
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS AND COMPUTING

Citation Formats

Guo, Boyun, Duguid, Andrew, and Nygaard, Ronar. Statistical Analysis of CO2 Exposed Wells to Predict Long Term Leakage through the Development of an Integrated Neural-Genetic Algorithm. United States: N. p., 2017. Web. doi:10.2172/1373948.
Guo, Boyun, Duguid, Andrew, & Nygaard, Ronar. Statistical Analysis of CO2 Exposed Wells to Predict Long Term Leakage through the Development of an Integrated Neural-Genetic Algorithm. United States. doi:10.2172/1373948.
Guo, Boyun, Duguid, Andrew, and Nygaard, Ronar. Sat . "Statistical Analysis of CO2 Exposed Wells to Predict Long Term Leakage through the Development of an Integrated Neural-Genetic Algorithm". United States. doi:10.2172/1373948. https://www.osti.gov/servlets/purl/1373948.
@article{osti_1373948,
title = {Statistical Analysis of CO2 Exposed Wells to Predict Long Term Leakage through the Development of an Integrated Neural-Genetic Algorithm},
author = {Guo, Boyun and Duguid, Andrew and Nygaard, Ronar},
abstractNote = {The objective of this project is to develop a computerized statistical model with the Integrated Neural-Genetic Algorithm (INGA) for predicting the probability of long-term leak of wells in CO2 sequestration operations. This object has been accomplished by conducting research in three phases: 1) data mining of CO2-explosed wells, 2) INGA computer model development, and 3) evaluation of the predictive performance of the computer model with data from field tests. Data mining was conducted for 510 wells in two CO2 sequestration projects in the Texas Gulf Coast region. They are the Hasting West field and Oyster Bayou field in the Southern Texas. Missing wellbore integrity data were estimated using an analytical and Finite Element Method (FEM) model. The INGA was first tested for performances of convergence and computing efficiency with the obtained data set of high dimension. It was concluded that the INGA can handle the gathered data set with good accuracy and reasonable computing time after a reduction of dimension with a grouping mechanism. A computerized statistical model with the INGA was then developed based on data pre-processing and grouping. Comprehensive training and testing of the model were carried out to ensure that the model is accurate and efficient enough for predicting the probability of long-term leak of wells in CO2 sequestration operations. The Cranfield in the southern Mississippi was select as the test site. Observation wells CFU31F2 and CFU31F3 were used for pressure-testing, formation-logging, and cement-sampling. Tools run in the wells include Isolation Scanner, Slim Cement Mapping Tool (SCMT), Cased Hole Formation Dynamics Tester (CHDT), and Mechanical Sidewall Coring Tool (MSCT). Analyses of the obtained data indicate no leak of CO2 cross the cap zone while it is evident that the well cement sheath was invaded by the CO2 from the storage zone. This observation is consistent with the result predicted by the INGA model which indicates the well has a CO2 leak-safe probability of 72%. This comparison implies that the developed INGA model is valid for future use in predicting well leak probability.},
doi = {10.2172/1373948},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Sat Aug 05 00:00:00 EDT 2017},
month = {Sat Aug 05 00:00:00 EDT 2017}
}

Technical Report:

Save / Share: