Development of a formal verification and validation database for the computational mass-flow simulator Titan2D
MetadataShow full item record
Titan2D is a numerical model developed for the purpose of simulating dry, cohesionless, granular avalanches over digitized elevation models (DEMs) of natural terrain. This includes certain classes of geological mass flows such as debris avalanches and landslides (where no interstitial fluid is present). Titan2D combines best-in-practice numerical methodologies for simulations of such flows with DEMs of natural terrain supported through a Geographical Information System (GIS) interface. As a computational model, the results produced by Titan inherently contain errors and uncertainties that can affect the accuracy of its solutions. Such artifacts arise throughout multiple levels of the modeling process such as in theoretical, experimental, and numerical (computational) domains. Verification and validation (V&V) are ongoing processes that attempt to provide a structured way to address such diverse sources of error and uncertainty. Verification addresses whether or not the computationally-implemented model is equivalent to the mathematical model on which it is based. Validation provides a determination of whether the computational model and its solution accurately represent the corresponding real-world phenomenon. In this paper, past efforts at validating Titan are reviewed and four new specific flow scenarios are modeled - an analytical Dam-Break scenario, a Channelized Flow experiment, a historic large-scale landslide (the 1903 Frank Slide in Alberta, Canada), and an inclined-plane, Deflected Sand Flow experiment. Each of these events has a "solution" (i.e. analytical, or empirical data) to which Titan's numerical results are compared. Through a series of such comparisons, it is hoped that a general verification and validation methodology emerges that can be used in the future to maintain a measure of Titan's accuracy regarding the simulation of various geophysical mass-flow events and related natural phenomena. Thus, the overriding goal of this report is two-fold. The first objective is to develop a comprehensive, reproducible database that enables one to identify, catalog, and understand the precise sources and nature of the errors, uncertainties, and limitations of the Titan2D code. Second, this database will be used as a test-suite , allowing the user/developer to compare specific flow solutions (case-studies) before and after any modifications to Titan to ensure proper functionality and self-consistency of the code.