11:15 AM - 11:30 AM
[MGI31-03] Estimation of depth profile of thermal conductivity of core samples from NanTroSEIZE using medical X-ray CT images
Keywords:IODP, Thrmal Conductivity, X-Ray CT Image, Digital Rock Physics
Thermal conductivity is one of the fundamental physical properties of rocks, and is important to estimate the temperature distribution in the subsurface for accurate understanding of seismic fault slip, understanding the formation conditions of underground resources, and predicting production rate. The thermal conductivity is usually measured by heating the core sample and measuring the temperature rise profile. However, with the recent development of X-ray CT (XCT) scanning technology and numerical computing environment, a method called "Digital Rock Physics (DRP)" has been proposed. This method loads microscopic internal structures of a sample from XCT images and simulates macroscopic physical phenomena by numerical calculations to estimate physical properties.
D/V Chikyu has acquired core samples for many scientific drilling expeditions since 2007. All core samples obtained during the International Ocean Discovery Program (IODP) expeditions by D/V Chikyu have been scanned by a medical X-ray CT scanner. Normally, this scan data is used for lithological observation, and the resolution is lower than that of images used for DRP, so it has not been used for estimating physical properties. However, if this data can be used for thermal conductivity estimation, it will be possible to obtain thermal conductivity with higher spatial resolution and for almost all core samples because the XCT images were routinely obtained for all core samples.
In this study, depth profiles of thermal conductivity were estimated using medical XCT images of core samples obtained at site C0002 during the Nankai Trough seismogenic zone experiments (NanTroSEIZE). Since the resolution of medical XCT images is lower, it is impossible to directly substitute physical properties to models after segmentation to separate pore and mineral phases. Therefore, the model of this study was created by "Segmentation-less method," which creates a continuous function between CT values and physical properties and substitutes different physical property values to each voxel. The core sample was divided into cubes of 2 cm square and models were created continuously in the depth direction using the calibration curve proposed in a previous study to calculate density and porosity to each voxel, and the mixing models of heat conduction obtain thermophysical properties.
The results show that the model is accurate compared to the measurements, although the trend varies with depth. On the other hand, the estimated prolife in the area affected by disturbances and cracks showed significant variation, resulting in inaccurate estimation. The methodology should be improved by removing the effect of disturbances and cracks in the near future.
D/V Chikyu has acquired core samples for many scientific drilling expeditions since 2007. All core samples obtained during the International Ocean Discovery Program (IODP) expeditions by D/V Chikyu have been scanned by a medical X-ray CT scanner. Normally, this scan data is used for lithological observation, and the resolution is lower than that of images used for DRP, so it has not been used for estimating physical properties. However, if this data can be used for thermal conductivity estimation, it will be possible to obtain thermal conductivity with higher spatial resolution and for almost all core samples because the XCT images were routinely obtained for all core samples.
In this study, depth profiles of thermal conductivity were estimated using medical XCT images of core samples obtained at site C0002 during the Nankai Trough seismogenic zone experiments (NanTroSEIZE). Since the resolution of medical XCT images is lower, it is impossible to directly substitute physical properties to models after segmentation to separate pore and mineral phases. Therefore, the model of this study was created by "Segmentation-less method," which creates a continuous function between CT values and physical properties and substitutes different physical property values to each voxel. The core sample was divided into cubes of 2 cm square and models were created continuously in the depth direction using the calibration curve proposed in a previous study to calculate density and porosity to each voxel, and the mixing models of heat conduction obtain thermophysical properties.
The results show that the model is accurate compared to the measurements, although the trend varies with depth. On the other hand, the estimated prolife in the area affected by disturbances and cracks showed significant variation, resulting in inaccurate estimation. The methodology should be improved by removing the effect of disturbances and cracks in the near future.