In part one of this study [1], a linkage was established between the pore structure and transport properties (sorptivity, diffusivity, and permeability) for a set of three common building materials. Using mercury intrusion porosimetry and scanning electron microscopy analysis in combination with two different types of computer models, computed transport properties based on the models compared favorably with those measured experimentally on the same suite of materials. In this present paper, this study is extended by directly analyzing the three-dimensional microstructure of two of the materials (bricks) as exemplified by X-ray microtomographic images obtained using beam line ID19 at the European Synchrotron Research Facilities (ESRF) in Grenoble, France. The two bricks examined are a lime silica brick formed by high pressure steam curing of a mixture of lime and silica, and a clinker brick which is a hard-burned clay brick. Both are typically used in the rain screens of cavity walls [1].
The 3-D images (256 pixels x 256 pixels x 256 pixels) are first processed to remove the random noise and the "circular ring" noise pattern often inherent in X-ray microtomographic images. The latter noise is characterized by a series of concentric rings radiating outward from the center of the sample in one of the image planes. After processing, the microstructures are binarized to match the overall porosity measured experimentally on the materials. For one of the bricks, the clinker brick, this is acomplished via a simple thresholding operation. For the lime silica brick, however, because the microstructure contains multi-scale features (pores) which are not resolvable using the X-ray tomographic equipment, a multi-scale approach to assigning porosity and computing transport properties is applied [1,2]. After binarization, 100 pixel x 100 pixel x 100 pixel or 200 pixel x 200 pixel x 200 pixel subsets of the resultant images are analyzed to compute diffusivities and permeabilities using previously developed techniques [1,3,4,5]. These computed values are compared to those previously measured experimentally [1]. It should be noted that the computational programs are not limited to transport properties, as shrinkage and elastic response can also be simulated on any 3-D binary image of a porous media [6].