3-DIMENSIONAL RECONSTRUCTION OF POROUS-MEDIA FROM SERIAL SECTION DATA

Software has been developed, with the aid of data for a Berea sandstone sample, to partition the irregular, interconnected pore space of a porous medium into its constituent pores. Output includes the sizes (volumes) of the individual pores, the sizes (area and radii of curvature) of the individual necks (windows) between the pores, the order in which the pores and necks are connected, and the degree of interconnectedness. Data are entered into the computer by digitizing a series of closely spaced serial sections of a pore cast of the sandstone. The pixels represent either rock matrix or pore space which are represented in binary format by ′0′ and ′1′, respectively. Each pixel is assumed to be the top surface of an associated volume element (voxel). The data array is analysed from seven orientations. For each orientation, a feature map is created which identifies the interconnected pore channels through the array. Then a set of potential necks is identified by locating local minima in the radius of curvature of pore space features along the pore channels. The seven sets of local minima are merged, overlaps among the minima are identified, and a consistent set of non‐overlapping necks is created by systematically deleting the larger overlapping minima. The necks are then ‘closed’ and the data set re‐analysed to identify each cluster of pore space voxels bound by rock matrix and ‘closed’ necks as a pore. 1990 Blackwell Science Ltd