<p>Carbon capture and storage (CCS) is a significant mitigation approach for decreasing carbon emissions, particularly in coal-dominated energy systems like South Africa, where coal combustion accounts for around 80% of national CO<sub>2</sub> emissions. The availability of credible datasets for subsurface characterisation, on the other hand, poses a significant problem for geological carbon storage. In other words, even with suitable geology, a lack of high-quality, comprehensive, and accessible subsurface data significantly limits the chance of identifying and quantifying viable storage locations, as well as predicting the long-term behaviour of stored CO<sub>2</sub>. Therefore, this study aims to evaluate the CO<sub>2</sub> storage potential of a depleted oil reservoir in the Bredasdorp Basin of South Africa. We used an integrated approach (petrophysics, machine learning, and petrography). This was achieved by adopting a robust methodological workflow that incorporated reports from thorough sedimentary petrography of thin-sectioned cored samples, coupled with wireline logs, which were subjected to petrophysical analysis using a combination of standard empirical relationships, six (6) different machine learning techniques, and core petrophysical analysis. As a result, the sandstones in the reservoir's cored intervals (sand AF) in both wells are compositionally and texturally quite similar and homogenous. The average volume of shale measured in the reservoir reveals that the sand is clean, with secondary porosity discovered in all of the porous sandstone samples following the breakdown of a precursor carbonate cement. When compared to empirical methodologies, the introduction of XG-Boost and Neural Network improved reservoir poroperm predictions, which were then calibrated using core data. As a result, the average porosity values of 15% and 16%, along with the reservoir permeability values of 193&#xa0;mD and 160&#xa0;mD, respectively, across the two wells indicate a good estimation of the CO<sub>2</sub> storage potential, significantly enhancing the performance of formation evaluation. The average fluid (water) saturation of the reservoir across both wells yields good values suitable for CO<sub>2</sub> injection, and the absence of reactive clays in the reservoir sheds light on sand AF's potential as a safe CO<sub>2</sub> sink in South Africa's Bredasdorp Basin. The findings from this work have shown that it is critical for South Africa to tap into its geologic CO<sub>2</sub> storage potential in hydrocarbon reservoirs, as it would be challenging to swiftly phase out fossil fuel energy sources in the near future and avoid an energy shortage crisis. Hence, the implementation of sustainable solutions is pertinent in order to mitigate climate change.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Petrophysics-based machine learning assessment of siliciclastic reservoir sediments for CO2 storage in the Bredasdorp Basin, Offshore, South Africa

  • Yinka Ayodeji Olayinka,
  • David Ubuara,
  • Ovie Emmanuel Eruteya,
  • Mimonitu Opuwari

摘要

Carbon capture and storage (CCS) is a significant mitigation approach for decreasing carbon emissions, particularly in coal-dominated energy systems like South Africa, where coal combustion accounts for around 80% of national CO2 emissions. The availability of credible datasets for subsurface characterisation, on the other hand, poses a significant problem for geological carbon storage. In other words, even with suitable geology, a lack of high-quality, comprehensive, and accessible subsurface data significantly limits the chance of identifying and quantifying viable storage locations, as well as predicting the long-term behaviour of stored CO2. Therefore, this study aims to evaluate the CO2 storage potential of a depleted oil reservoir in the Bredasdorp Basin of South Africa. We used an integrated approach (petrophysics, machine learning, and petrography). This was achieved by adopting a robust methodological workflow that incorporated reports from thorough sedimentary petrography of thin-sectioned cored samples, coupled with wireline logs, which were subjected to petrophysical analysis using a combination of standard empirical relationships, six (6) different machine learning techniques, and core petrophysical analysis. As a result, the sandstones in the reservoir's cored intervals (sand AF) in both wells are compositionally and texturally quite similar and homogenous. The average volume of shale measured in the reservoir reveals that the sand is clean, with secondary porosity discovered in all of the porous sandstone samples following the breakdown of a precursor carbonate cement. When compared to empirical methodologies, the introduction of XG-Boost and Neural Network improved reservoir poroperm predictions, which were then calibrated using core data. As a result, the average porosity values of 15% and 16%, along with the reservoir permeability values of 193 mD and 160 mD, respectively, across the two wells indicate a good estimation of the CO2 storage potential, significantly enhancing the performance of formation evaluation. The average fluid (water) saturation of the reservoir across both wells yields good values suitable for CO2 injection, and the absence of reactive clays in the reservoir sheds light on sand AF's potential as a safe CO2 sink in South Africa's Bredasdorp Basin. The findings from this work have shown that it is critical for South Africa to tap into its geologic CO2 storage potential in hydrocarbon reservoirs, as it would be challenging to swiftly phase out fossil fuel energy sources in the near future and avoid an energy shortage crisis. Hence, the implementation of sustainable solutions is pertinent in order to mitigate climate change.