This paper has reviewed the overall procedure for acquiring aggregate shape data from x-ray computed tomography (CT), including sample preparation, 2-D and 3-D image analysis, 3-D particle reconstruction from 2-D slices, and error correction to eliminate artificially mis-shaped particles. VRML images were shown to be easily generated for each particle, and useful for qualitative examination and comparison to real images. Reference rocks were used for quality control of the whole process, from x-ray CT through image analysis to spherical harmonic mathematical analysis. Results were displayed for four different kinds of aggregates and three reference rocks, illustrating the sorts of shape analysis possible using complete 3-D shape information that this process gives.
There is a wide array of aggregates used in the U.S. and an even greater range of material worldwide. It is known from sedimentary petrology that characteristic shapes can often be qualitatively predicted from knowledge of geological deposits and geomorphological processes. However, these methods cannot give precise numerical information. But we cannot image them all. A research goal is, using information obtained from a statistical sample of aggregates (100's to 1000's), to be able to statistically and realistically generate aggregates based on morphological descriptions [22]. This goal requires a solid and detailed numerical foundation on which to base such statistical predictions. Hence, there is a need for imaging more kinds of aggregates, in order to expand our database and to develop correlations between spherical harmonic coefficients and physical shape as well as standard shape descriptors such as ASTM D-4791 [14]. Examining mathematical differences between crushed and naturally rounded aggregates would probably be instructive. The large amount of data that the spherical harmonic coefficients represent open up many possibilities for mathematical analysis of aggregate shape. A few possibilities were discussed in this paper, but many more need to be studied [11].
Expanding the aggregate database in the Virtual Cement and Concrete Testing Laboratory (VCCTL) will also allow better prediction of a wider range of concrete mixes. The VCCTL is an integrated software package for predicting the properties of concrete from knowledge of the basic ingredients, hydration chemistry, and curing conditions [23]. Having basically complete 3-D shape information for many aggregates will give us an unprecedented ability to realistically model the structure and predict the properties of concrete.