Publications

Abstract not provided.

This paper presents the conceptual framework that is being used to define quantification of margins and uncertainties (QMU) for application in the nuclear weapons (NW) work conducted at Sandia National Laboratories. The conceptual framework addresses the margins and uncertainties throughout the NW life cycle and includes the definition of terms related to QMU and to figures of merit. Potential applications of QMU consist of analyses based on physical data and on modeling and simulation. Appendix A provides general guidelines for addressing cases in which significant and relevant physical data are available for QMU analysis. Appendix B gives the specific guidance that was used to conduct QMU analyses in cycle 12 of the annual assessment process. Appendix C offers general guidelines for addressing cases in which appropriate models are available for use in QMU analysis. Appendix D contains an example that highlights the consequences of different treatments of uncertainty in model-based QMU analyses.

Abstract not provided.

Experienced experimentalists have gone through the process of attempting to identify a final set of modal parameters from several different sets of extracted parameters. Usually, this is done by visually examining the mode shapes. With the advent of automated modal parameter extraction algorithms such as SMAC (Synthesize Modes and Correlate), very accurate extractions can be made to high frequencies. However, this process may generate several hundred modes that then must be consolidated into a final set of modal information. This has motivated the authors to generate a set of tools to speed the process of consolidating modal parameters by mathematical (instead of visual) means. These tools help quickly identify the best modal parameter extraction associated with several extractions of the same mode. The tools also indicate how many different modes have been extracted in a nominal frequency range and from which references. The mathematics are presented to achieve the best modal extraction of multiple modes at the same nominal frequency. Improvements in the SMAC graphical user interface and database are discussed that speed and improve the entire extraction process.

An algorithm known as SMAC (Synthesize Modes And Correlate), based on principles of modal filtering, has been in development for a few years. The new capabilities of the automated version are demonstrated on test data from a complex shell/payload system. Examples of extractions from impact and shaker data are shown. The automated algorithm extracts 30 to 50 modes in the bandwidth from each column of the frequency response function matrix. Examples of the synthesized Mode Indicator Functions (MIFs) compared with the actual MIFs show the accuracy of the technique. A data set for one input and 170 accelerometer outputs can typically be reduced in an hour. Application to a test with some complex modes is also demonstrated.

As model validation techniques gain more acceptance and increase in power, the demands on the modal parameter extractions increase. The estimation accuracy, the number of modes desired, and the data reduction efficiency are required features. An algorithm known as SMAC (Synthesize Modes And Correlate), based on principles of modal filtering, has been in development for a few years. SMAC has now been extended in two main areas. First, it has now been automated. Second, it has been extended to fit complex modes as well as real modes. These extensions have enhanced the power of modal extraction so that, typically, the analyst needs to manually fit only 10 percent of the modes in the desired bandwidth, whereas the automated routines will fit 90 percent of the modes. SMAC could be successfully automated because it generally does not produce computational roots.

This paper investigates the use of artificial neural networks (ANNs) to identify damage in mechanical systems. Two probabilistic neural networks (PNNs) are developed and used to judge whether or not damage has occurred in a specific mechanical system, based on experimental measurements. The first PNN is a classical type that casts Bayesian decision analysis into an ANN framework, it uses exemplars measured from the undamaged and damaged system to establish whether system response measurements of unknown origin come from the former class (undamaged) or the latter class (damaged). The second PNN establishes the character of the undamaged system in terms of a kernel density estimator of measures of system response; when presented with system response measures of unknown origin, it makes a probabilistic judgment whether or not the data come from the undamaged population. The physical system used to carry out the experiments is an aerospace system component, and the environment used to excite the system is a stationary random vibration. The results of damage identification experiments are presented along with conclusions rating the effectiveness of the approaches.

We present a software environment integrating analysis and test based models to support optimal modal test design through a Virtual Environment for Test Optimization (VETO). The VETO assists analysis and test engineers in maximizing the value of each modal test. It is particularly advantageous for structural dynamics model reconciliation applications. The VETO enables an engineer to interact with a finite element model of a test object to optimally place sensors and exciters and to investigate the selection of-data acquisition parameters needed to conduct a complete modal survey. Additionally, the user can evaluate the use of different types of instrumentation such as filters, amplifiers and transducers for which models are available in the VETO. The dynamic response of most of the virtual instruments (including the device under test) are modeled in the state space domain. Design of modal excitation levels and appropriate test instrumentation are facilitated by the VETO`s ability to simulate such features as unmeasured external inputs, A/D quantization effects, and electronic noise. Measures of the quality of the experimental design, including the Modal Assurance Criterion, and the Normal Mode indicator Function are available. The VETO also integrates tools such as Effective Independence and minamac to assist in selection of optimal sensor locations. The software is designed about three distinct modules: (1) a main controller and GUI written in C++, (2) a visualization model, taken from FEAVR, running under AVS, and (3) a state space model and time integration module, built in SIMULINK. These modules are designed to run as separate processes on interconnected machines. MATLAB`s external interface library is used to provide transparent, bidirectional communication between the controlling program and the computational engine where all the time integration is performed.