Uncertainty Quantification in Model Verification & Validation as Applied to Large Scale Historic Masonry Monuments
|Date||05.07.2013 – 11:00 › 12:00|
Sez Atamturktur, Assistant Professor
Glenn Department of Civil Engineering, Clemson University, South Carolina, USA
|Category||Conferences - Seminars|
This presentation will focus on the Verification and Validation (V&V) of numerical models for establishing confidence in model predictions with a specific focus on large scale masonry monuments. Through an overview of various case study applications, the presentation will discuss relevant phases of a V&V process: numerical model development, field experimentation and model verification & validation. The main goals of this presentation are to discuss where modeling errors and uncertainty originate from, and present a V&V approach to obtain model predictions that are statistically consistent with their respective measurements. Our V&V approach considers all major sources of errors and uncertainty that originate from numerical solutions of differential equations (numerical uncertainty), imprecise model input parameter values (parameter uncertainty), incomplete definitions of underlying physics due to assumptions and idealizations (bias error) and variability in measurements (experimental uncertainty). In the presentation, a brief overview of best-practices for designing calibration experiments will be provided with a particular emphasis on in situ vibration measurements. Specifically, an approach for optimal placement of sensors utilizing an information-based criterion will be discussed to not only reduce the cost and time demands of on-site testing but also the amount of measurement data requiring post-processing. The presentation will conclude with a discussion on the role of selected model form (i.e. macro versus micro modeling of masonry assembly) in regards to the robustness and fidelity of model predictions.