This report addresses the applicability of Singular Value Decomposition (SVD) for answering facility design and radiographic reconstruction questions for the Advanced Hydrotest Facility (AHF). SVD analysis provides an richly informative linear decomposition of any projection matrix (which is equivalent to experiment design). A library of low-dimensional test objects and projection matrices is used to quantitatively illustrate the types of information readily available to facility designers and users. In particular, it is shown that the following questions can be addressed in a systematic way: What test object parameterizations ensure optimal use of data? How many detector views are necessary for obtaining reconstructions of sufficient quality? How can detector views be optimally placed? What are the best methods for reconstructing images from noisy data? These questions are ultimately tied to stockpile requirements and the SVD provides a natural tool for examining this relationship. Specific quantitiative answers for the AHF depend upon stockpile and experimental considerations yet to be exactly determined. Thus, the numerical results of this report are not indicators of AHF performance and object-reconstruction capabilities. However, they do demonstrate how concrete numerical results are readily obtainable given experimental performance criteria and stockpile metrics. In addition, this report outlines some details of promising research avenues including nonlinear projection noise stability analysis, modified reconstruction subspaces, total variation minimization reconstruction enhancement, and the use of dynamical constraints.

LA-UR-01-6534