利用多解析度X光電腦斷層掃描與統計指標進行砂岩滲透性之研究

Abstract

在全球氣候變遷的背景下，地質封存被視為減少大氣中二氧化碳濃度的有效途徑。為提升二氧化碳在地下環境傳輸過程的理解，建立岩芯尺度之滲透性模型並據此進行數值模擬至關重要。然而，傳統滲透性測量透過採取岩芯栓進行實驗，通常僅能給出單一樣本的平均值且樣本量有限；醫學電腦斷層掃描配合灌注實驗可呈現岩芯尺度之滲透性分布但解析度不足；微米級電腦斷層掃描能解析孔隙資訊並可透過模擬方法計算滲透性，卻受限於視場，無法描述公分尺度以上的異質性。因此，本研究整合多解析度電腦斷層影像，藉此同時呈現岩芯尺度的異質性與孔隙尺度的特徵，以改善岩芯尺度滲透性模型估計的準確性。為此，本研究選取桂竹林層大埔段、南莊層與卓蘭層的四個岩芯栓樣本，於5.0、22.3 與 68.9 微米解析度下獲取電腦斷層掃描資料，透過閾值分割建立其二值化影像與孔隙網路模型以取得孔隙率、孔隙與孔頸尺寸、連通度與模擬滲透性，並與實驗測量對照評估其可靠性。結果顯示5.0微米影像的模擬滲透性與實驗測量相近，而22.3與68.9微米影像受點擴散與部分體積效應影響導致模擬滲透性結果不佳，但其孔頸尺寸仍與實驗滲透性呈高度正相關。在此基礎上，為降低閾值分割不確定性，本研究提出下偏標準差作為影像訊號的統計指標，以描述孔隙尺寸的變化，結果發現其在各解析度下均與孔隙尺寸、孔頸尺寸及實驗滲透性之關係呈現顯著正相關，顯示下偏標準差可作為跨解析度的滲透性指標。進一步以南莊層之非均質樣本進行跨解析度驗證，結果顯示，採用下偏標準差於68.9微米解析度下推估之滲透性分布，與22.3微米解析度結果高度一致。鑑於68.9微米解析度可應用於約十公分口徑之全岩芯電腦斷層掃描，本研究之作法可拓展至全岩芯影像以建立岩芯尺度滲透性分布，可更忠實呈現次岩芯尺度的材料異質性，並提升岩芯尺度二氧化碳傳輸模擬之可信度，有助於場址注入規劃、封存成效評估與長期安全性之模型校正與決策。

In response to escalating climate challenges, geological sequestration is regarded as an effective means of reducing atmospheric CO₂ concentrations. To advance the understanding of CO₂ transport processes in the subsurface, it is essential to construct core-scale permeability models and conduct numerical simulations. However, conventional permeability measurements conducted on core plugs provide only an average value for a single specimen and are limited in sample amount. Medical CT scans combined with core flooding experiments reveal core-scale permeability distributions but lack sufficient resolution. Micro-CT resolves pore-scale features and enables permeability simulation, but its limited field of view hinders characterization of heterogeneity beyond a few centimeters. Accordingly, this study integrates multi-resolution CT imaging to simultaneously capture core-scale heterogeneity and pore-scale features, thereby improving the accuracy of core-scale permeability models. To this end, four core-plug samples from the Kueichulin Formation (Dapu Member), the Nanchuang Formation, and the Cholan Formation were selected and scanned at resolutions of 5.0 μm, 22.3 μm, and 68.9 μm. Thresholded binary images and pore network models (PNM) were constructed at each resolution to obtain porosity, pore size, throat size, connectivity, and simulated permeability, which were then compared with experimental measurements to evaluate their reliability. The results show that simulated permeability derived from the 5.0 μm images is consistent with experimental measurements, whereas the 22.3 μm and 68.9 μm images are affected by point-spread and partial volume effects, leading to poor simulated permeability. Nevertheless, throat size remains highly correlated with experimental permeability at three resolutions. Building on this, to reduce threshold segmentation uncertainty, this study introduces the lower partial standard deviation (LPSD) as a grayscale-based indicator of pore-size variation. The results demonstrate that LPSD exhibits significant positive correlations with pore size, throat size, and experimental permeability across all resolutions, indicating its potential as a cross-resolution permeability proxy. Further cross-resolution validation on a heterogeneous sample from the Nanchuang Formation shows that permeability distributions estimated using LPSD at 68.9 μm and 22.3 μm are highly consistent. Given that 68.9 μm resolution can be applied to whole-core CT scans with diameters of ~10 cm, the proposed approach thus establishes core-scale permeability distributions. This enables a more faithful representation of sub-core-scale heterogeneity and enhances the reliability of core-scale CO₂ transport simulations, thereby supporting injection strategies, storage performance evaluation, and long-term safety assessment for geological storage projects.