Previous LADSS Programs

2024

Projects and Teams

Design of a Tuned Vibration Absorber with Friction Contact for High Location Ratio Application

Mentors: Sean Kelly (E-14), Thomas Thompson (E-14), Keegan Moore (Georgia Institute of Technology)
Students: Matthew Hancock, Nathan Lovell, Yusuf Shehata

Tuned vibration absorbers (TVAs) are passive devices commonly used to reduce the vibration amplitude of a host structure subject to harmonic excitations. These devices are designed so that vibration energy ﬂows from the host structure to the TVA. To dissipate this energy, common TVAs use linear damping mechanisms. However, nonlinear energy dissipation methods such as friction-enhanced TVAs have recently been explored. Additionally, it is desirable to place a TVA on a host structure where large vibration amplitudes occur, but this is often not feasible in practice. The ratio of the amplitudes in a desirable high-vibration-amplitude area to a low-vibration-amplitude area is referred to as the location ratio (LR). Careful tuning of the TVA is critical to ensure absorption eﬀectiveness especially when the TVA is placed in a high-LR. This paper presents a case study of a novel TVA design that is optimized for high- LR applications and utilizes nonlinear friction contacts to dissipate energy. The host structure analyzed is the Box Assembly (BA), an experimental structure previously analyzed for vibration testing. A ﬁnite- element model of the BA, validated by experimental stepped sine tests, is considered, and a target mode of the BA is selected for TVA design. Simulations are performed to illustrate the eﬀectiveness of the TVA in a high-LR application by varying linear system parameters. The eﬀects of TVA nonlinear friction damping on the vibration amplitude reduction is also investigated. From these simulations, design considerations are suggested.

Crack detection and localization optimization in Bi2Te3 wafers

Mentors: John Greenhall (MPA-11), Milo Prisbrey (MPA-11), Nick Lieven (University of Bristol)
Students: Angel Bravo, Thomas Magee, Sterling Rawls

Evaluating the Effect of Shaker Placement Optimization Priorities on Multi-Axis Test Results

Mentors: Shannon Danforth (E-14), Brittany Ouellette (E-14), John Schultze (E-14), Jim De Clerck (Michigan Tech University)
Students: Risto Djishev, Kieran Elrod, Connor Tasik

Multi-axis vibration testing is capable of emulating ﬁeld environments more accurately than traditional single-axis testing. However, multiple shakers introduce challenges in test setups and execution, which can be addressed with optimization algorithms. One example is to select shaker locations that minimize error at a control location (objective) while keeping the shaker output force lower than a speciﬁed threshold (constraint). The solution given by the algorithm is aﬀected by many variables, including the choice of cost and constraints, metrics used for formulating the cost and constraints, and model ﬁdelity and calibration. Implementation of optimal shaker placements in the laboratory will result in diﬀerent boundary conditions, shaker and sensor attachment method, and shaker control which aﬀects how the optimization criteria are reﬂected in the test results. This case study investigates the challenges and beneﬁts of optimizing shaker location by formulating two diﬀerent optimization problems, implementing each suggested shaker conﬁguration in laboratory tests, and comparing results to speciﬁed objectives and constraints.

Capture of Thin Film Vibromorphology using Laser Vibrometry and Frequency-Domain Viscoelastic Vibroacoustics

Mentors: Eric Bryant (EES-17), Jeff Tippmann (ISR-6), WaiChing Sun (New York University)
Students: Kent Eng, Steven Kao, Natalie White

Selection of Optimal Reduced-Order Model of a Device Under Test Using Bayesian Classification

Mentors: Colin Haynes (E-14), Thomas Thompson (E-14), Laura Redmond (Clemson University)
Students: Chris Garcia, Jessica Heinlen, Devon Scheg

To predict whether a shaker system has the capability to perform a specific environmental test, a coupled system model of the shaker system and the device under test (DUT) is often used. Using a high-fidelity representation of the DUT within this coupled model would be the most accurate way to determine if the shaker can adequately excite a high degree of freedom (DOF) DUT to a broadband set of frequencies within its voltage and force output limits. However, this model is computationally expensive to run repeatably. To limit the computation time, reduced-order models (ROM) of the DUT can be implemented.

This paper analyzes ROMs of various fidelity using a Bayesian Classifier to determine what ROM is most like the available test data and determine what parameters are most important to maintain accuracy while reducing computation time. By using this method, trends of parameter selection are identified to be used to guide future ROMs for a DUT.

Extrapolating Dynamic Transfer Functions from Multi-Input Multi- Output Vibration Testing and Simulation

Mentors: Sandra Zimmerman (ISR-5), Eli Dawson (E-14), John Blessinger (W-13), Jul Davis (University of Southern Indiana)
Students: Elijah Perez, Ethan Regula, Carson Wynn

Finding transfer functions between and applied force and a measured output is well defined for many types of dynamic systems. These transfer functions, which assume a force input, can have many names depending on the output they measure, such as receptance for displacement, mobility for velocity, or accelerance for acceleration. For single-input linear systems, finding the transfer path between two outputs is a well-understood calculation. These techniques become more complex when dealing with multi-input multi-output tests. While these vibration tests are more complicated, they provide a more complete understanding of the system being analyzed. These complex systems have well-defined methods for obtaining transfer functions if a forcing function is known. However, when the input force isn’t known, the methods to find the transfer path between two locations becomes much more complicated.

This work explores methods for finding these transfer paths, independent of forcing function, using accelerations. The tests and validations are done using the Box Assembly, a common structure used in dynamics testing. Experimentally measured and finite-element approximated transfer functions are used to create a transmissibility matrix. This matrix allows for estimation of unknown accelerations based on measured accelerations at different locations on the box assembly. Another application of the methods derived in this work allow for the location of a force to be calculated from the transfer paths found.

Characterization of Rayleigh-Taylor Instabilities Induced by Volumetric Energy Deposition

Mentors: Adam Wachtor (EI), Joseph Kerwin (E-14), Ricardo Mejia (Michigan State University)
Students: Clara Bender, Anna Cardall, Fabian Rodriguez

Internal confinement fusion (ICF) exhibits Rayleigh-Taylor (RT) instabilities during the compression of target fuel capsules, leading to uneven heating and a reduction of efficiency. While computational models for RT instability in the context of ICF exist, lab-scale experimental work is needed to validate the assumptions and results of these models. Much experimental work investigating RT instability has been performed, but few studies involve the condition of volumetric heating or the resultant fluid density variation that is present in ICF. This work demonstrates RT instability induced by volumetric energy deposition (VED), which caused buoyancy-driven mixing due to density stratification inversion.

Experimentation involved a vessel with an initially stable system of two density-stratified immiscible fluids. Due to its higher dielectric loss, the denser fluid at the bottom of the vessel experienced preferential heating and thermal expansion by microwave VED. Buoyancy-driven mixing was then induced when the density stratification was inverted. Since the refractive indexes of the fluids are different, the phenomenon was visualized with shadowgraph videography. Once the mixing process was recorded, images were processed with methods such as edge detection, segmentation, filtering, pattern matching, and motion tracking techniques. Image processing was used to unveil the presence of morphological features of the flow such as spikes and bubbles, and to characterize the evolution and dynamic interaction of the fluids. Finally, statistical analysis was carried out to identify turbulence descriptors relevant to RT mixing.

2023

Projects and Teams

A Characterization of the Uncertainty in Force-Control Testing for Aerospace Applications

Mentors: Garrison Flynn, Colin Haynes, Keegan Moore (University of Nebraska – Lincoln)
Students: Katie Hart, Eddie Lewis, Shanell Sinclair

In many aerospace applications, it is impossible to measure the forces, both aerodynamic and structure-borne, applied to a structure in its service environment. Furthermore, collecting service environment data requires a modified flight structure which accommodates telemetry equipment. Due to the difference in the mechanical impedance of the shaker table and payload, the test structure experiences different loads in the test environment which can lead to over- and under-testing of the structure. Consequently, force-control testing may be a more desirable testing procedure as it more accurately reflects the service environment and eliminates the adverse effects caused by the differences in mechanical impedance. As part of this procedure, a numerical model is required to derive the test forces to replace the acceleration control. Determining these test forces from the model will introduce uncertainty into the flight structure’s response profile. This paper aims to quantify the uncertainty in the determination of the test forces used in the force-controlled test and the uncertainty in the test structure’s response.

Computerized Acoustic Resonance Crack Identification in Thermoelectric Bi2Te3 Wafers

Mentors: John Greenhall, Milo Prisbrey, Bart Raeymaekers (Virginia Tech)
Students: Ruth Hammond, Lexy Murphy, Lindsay Wright

Thermoelectric materials, including bismuth telluride (Bi2Te3), possess the remarkable ability to convert heat into electricity, or, conversely, convert electricity into heat, mediated by the Peltier effect. These materials find application in various engineering products, such as powering space exploration vehicles and enabling efficient wafer cooling in electrical systems. However, this material is susceptible to microcracking, reducing the reliability and usefulness of these materials. The current industry practice involves laborious manual inspections of each wafer to identify any defect, potentially increasing preexisting cracks and damaging otherwise useable wafers. This project focused on expediting crack location through nondestructive means. This experimental setup employed an acoustic resonance to excite the wafer to reveal existing cracks. The “excited” wafer was then scanned with a laser Doppler vibrometer (LDV) to visually pinpoint fracture locations. We then utilize an algorithm based on image processing and machine learning (ML) to analyze the obtained data scans to locate microcracks on the wafer. This results in an accurate, automated method for defect detection in thermoelectric wafers that could also be extended to detecting microcracks in other devices such as silicon microchips and micro electro-mechanical systems (MEMS).

Evaluating Vibration Controller Performance in Virtual and Hardware Tests

Mentors: Shannon Danforth, Brittany Ouellette, John Schultze
Students: Tessa Lytle, Wyatt Saeger, Aiden Tombuelt

Multi-axis testing has gained popularity in the dynamic environments testing community due to its potential to more accurately recreate observed field conditions and reduce test durations. However, the increased sensors and excitation sources necessary for testing in a multi-axis setting add complexity to planning and execution. Two important components of a successful multi-axis test are (1) the ability to run accurate virtual tests to determine the optimal shaker and sensor configuration and (2) a vibration controller that can produce the shaker forces necessary to match the test specification at control locations on the test article. This project aims to evaluate the control capabilities within virtual tests and between virtual and hardware tests for the base section of a Box Assembly with Removable Component (BARC). The Rattlesnake Vibration Controller software developed at Sandia National Laboratories is used to conduct the virtual and hardware tests in this study, and a finite element model (FEM) of the BARC base forms the model for the virtual tests. It is expected that the virtual test’s ability to predict hardware test results will depend on model fidelity, control locations, boundary conditions, and the test specification characteristics. By analyzing how each of these factors contributes to a virtual-to-hardware test pipeline, this study facilitates the development of new vibration control strategies and test planning optimization frameworks. 

Adaptive Radio Frequency Target Localization

Mentors: Zig Hampel-Arias, Jeff Tippmann
Students: Anthony Petrakian, Parker Segelhorst, Abby Smith

Mobile radio frequency (RF) target localization using signal characterization is a developing field with a wide variety of applications including structural health monitoring, military threat localization, and search and rescue. This is typically done using a greedy approach, where a sensor is moved closer to the target after each measurement. However, this does not allow for other constraints, such as maintaining a fixed distance from the target or optimization of energy consumption, for example. To permit such constraints, machine learning techniques such as Bayesian state estimation, reinforcement learning, and particle filters have been used to localize dynamic targets with line-of-sight observations, and static targets with RF observations. A recent study simulated localization of static and dynamic targets through RF signal characterization without the need for line-of-sight observation. This was done by employing a Partially Observable Markov Decision Process (POMDP), which was constructed by defining a state for the system that contained localization information for the sensor and the target, an action space to allow for movement of the sensor, and a reward function to allow for additional constraints for the model and to dictate specific behaviors. The purpose of this work is to build upon this prior study by applying previously employed solution strategies with real world data to measure the accuracy of the model. This will be done by starting with a static sensor and target, and gradually removing constraints until the sensor and target are both dynamic. From the information gathered, solution strategies can be adjusted after each test to better reflect real world environments.

Quantitative Comparison of Vibration Testing Methods

Mentors: Peter Fickenwirth, Thomas Thompson, Sandra Zimmerman
Students: Tharwat Elkabani, Celvi Lisy, Gerrit Vander Wiel

Dynamic environmental testing is an essential part of qualifying aerospace structures and components for transportation and flight. Traditional environmental testing subjects test articles to three single input single output tests, orthogonally exciting test specifications to approximate a three-dimensional field input to the article. However, altered boundary conditions between the laboratory and the field result in differing environments imparted onto the article. Further, uncontrolled motion out-of-axis from the direction of excitation still imparts energy onto the test-article. The dynamic environmental testing community has begun to explore multi-axis testing to better approximate field environments and save time on testing. Challenges in single axis tests includes ignoring certain stress states and failure modes that should otherwise exist in the test. Multi-input multi-output testing is shown to reduce over and under testing, test times, and experimental setup times. 

Baseline field data is used as an input for random environmental testing in both single and multi-axis tests. The structure analyzed in vibration tests was the Box Assembly with Removable Component (BARC) developed at Sandia National Laboratory and Kansas City National Security Campus, as a benchmark for nonlinear dynamic testing. Random field data was gathered by transporting the BARC structure in a vehicle across bumpy roads. Metrics must be developed to compare single-axis and multi-axis tests to quantitatively represent advantages and disadvantages seen in single-axis and multi-axis for a given experimental setup. This project seeks to develop metrics that effectively and quantitatively compare the fidelity of dynamic environmental testing methods. Potential metrics for comparing tests include fatigue damage spectrum (FDS), root mean square (RMS), and power spectral density (PSD). These metrics should encompass the success of the test at control locations as well as uncontrolled locations and in different frequency ranges tested.

Identification of Bird Species in Large Multi-channel Data Streams Using Distributed Acoustic Sensing

Birds are common indicators for monitoring ecological health, due to their ubiquity across habitat types and their greater sensitivity to environmental changes compared to other vertebrates. The use of acoustic recordings to sample bird communities has recently been established as an effective collection method for estimating bird species richness and abundance, as well as for identifying indicator species when compared to point counts (the traditional method of collecting bird tallies with an observer). Distributed acoustic sensing (DAS) is an emerging technology that utilizes Rayleigh backscattering measurements to assess how much strain is present at different points in a singular optical fiber, and acts as a multidimensional, robust, high sampling rate data collection method. It has gained popularity due to its vast applications in seismic wave detection, pipeline monitoring, and traffic monitoring. Unlike typical audio recording configurations, the optical fiber is easily installed and requires only a single DAS interrogator to collect kilometers of acoustic and spatial data, requiring no power at the actual locations of sensors. Given the large quantity of data output by DAS, automatic detection algorithms are necessary to filter and identify acoustic signatures. By developing a signal matching algorithm, we show the feasibility of using DAS to monitor birds of interest for environmental monitorization. 

Generating and Quantifying Initial Conditions for Volumetric Energy Deposition Driven Rayleigh-Taylor Instability Experiments

Mentors: Adam Wachtor, Brandon Wilson
Students: Cade Engen, Philip Root

Experimental validation of fluid dynamic simulations requires careful control and characterization of the initial flow conditions, such as characteristic length scales, energy spectra, and fluid properties. This need is especially pertinent for transient flows and those developing from instabilities. In this work, a novel Rayleigh-Taylor instability experiment was designed and conducted to study the mixing layer growth rate of a two-fluid interface when the change in the magnitude of the density gradient between the immiscible fluids was controlled through preferential heating of one fluid via volumetric energy deposition. The fluids were selected such that the system was initially Rayleigh-Taylor stable, but transitioned through the instability point through preferential microwave heating of one of the fluids. In order to control the morphology of the two-fluid interface, mechanisms to create Faraday waves at the interface were investigated. It was the perturbations of the interface that caused a misalignment of the density gradient and pressure gradient that drove instability growth. Volumetric energy deposition continued throughout the entire experiment, continuously increasing the density difference between the two fluids from the initial onset of the instability. Shadowgraphy of the fluid system was captured to quantify the resultant effects of the mixing layer growth from changes to the interface morphology, fluid properties, and microwave heating rate.

2022

Projects

Characterizing the Dynamic Response of a Foam-Based Testbed with Material, Geometric, and Experimental Uncertainties

Mentors: Thomas Roberts, Samantha Ceballes, Scott Ouellette
Students: Tariq Abdul-Quddoos, Patrick Lee, Cole Zemelka

Closed-cell polymer foams are commonly employed as support structures to absorb shock and vibration in mechanical systems. Engineering analysts responsible for system designs that incorporate these foams must understand the effects that intrinsic and extrinsic conditions have on their dynamic responses. Parameters intrinsic at the system level, such as preloading and material properties along with extrinsic environmental parameters, such as forcing energy and frequency, have the potential to drive the system into non-linear or chaotic regimes. A suite of simulation-based studies is performed utilizing finite element (FE) analysis to investigate both intrinsic and extrinsic model parameters to understand such effects on the non-linear system dynamics. A high-fidelity FE model of the mass-foam testbed needs to be developed to perform two tasks – implicitly determine stress states from pre-compression in the foam and explicitly solve for the system’s response when subject to various dynamic inputs. Using prior knowledge of stochastic quantities in the model, input parameter distributions will be generated to quantify the influences on the system’s dynamic response behavior. An existing testbed consisting of a mass suspended by two pieces of closed-cell polymer foam will be used to perform various experiments for model validation. This project aims to explore the parameter spaces to predict, validate, and confidently bound the non-linear dynamic behaviors of a system.

Designing Accelerated Vibration Tests using Model-Based Equivalent Damage Prediction

Mentors: Thomas Thompson, Garrison Flynn, Keegan Moore (University of Nebraska - Lincoln)
Students: Taylor Kinnard, Davis McMullan, Katherine Pane

The use of vibration testing to complete qualification of critical components is important for a wide variety of industries to understand the life cycles of their products in operational environments. A common problem with testing components to failure is the time and cost associated with mimicking the full life of a part, creating a need for shorter-duration testing that provides comparable life cycle information. The most common methods that accelerate damage tests use Miner’s Rule, an equation that sums damage percentages caused by varying stress amplitudes. The aim of using Miner’s Rule in damage analysis is to find a shorter-duration test cycle that will provide equivalent damage to the part’s real-world environment. This method has demonstrated accuracy under constant amplitude loading but loses reliability under variable amplitude loadings due to its lack of regard toward the loading sequence. Furthermore, translating from a stress-cycle (SN) curve to design amplitudes for testing requires system knowledge. Finally, the entire process of damage equivalence for additively manufactured (AM) parts is minimally explored in current research. This study seeks to improve the quality of test acceleration by utilizing models of the system under test to not only provide a method for faster, more accurate equivalent damage analysis, but also to fill a void of a lack of information regarding test compression of AM parts. To do this, AM specimens, designed with a failure point under a complex stress history, are evaluated. First, parts are modeled using finite element analysis and various vibration loads are simulated. Next, predicted stress responses are used to design test inputs for a shaker test within a desired time to failure. Finally, relationships between stress and cycles to failure were compared between the experimental and theoretical models to benchmark the accuracy of test design using modeled stress predictions and Miner’s rule. Test results of the experimental setup and simulated environment were compared to evaluate the accuracy of Miner’s rule in equivalent damage analysis, as well as test accuracy of SN curves for designing test of AM parts.

A Simplified Finite Element Joint Model Updated with Experimental Modal Features

Mentors: Nicholas Lieven (University of Bristol), Christopher Johnson, Manuel Vega
Students: Jonathan Black, Skylar Callis, Aaron Feizy

Finite element modeling (FEM) and analysis (FEA) are commonly employed for structural design evaluation and iteration. With current technology, full-fidelity modeling of larger assemblies is often computationally prohibitive, requiring simplification of the model's complex features. One such feature that is frequently simplified is the bolted joint, which is ubiquitous in engineering structures. However, approximate methods for modeling joints introduce inaccuracies. To better understand this model-induced error, this paper explores a novel computationally efficient method for the dynamic modeling of bolted connections. This method was applied to finite element modeling of bolted connections in a four-story structure. Experimental modal analysis was conducted to validate the joint modeling method. Bayesian model calibration was used to quantify the model parameter uncertainties and update geometric and material properties.

Additively Manufactured Component Characterization by Machine Learning from Resonance Inspection Techniques

Mentors: TJ Ulrich, Parisa Shokouhi (Pennsylvania State University)
Students: Stephanie Gonzalez, Sierra Horangic, Joseph Lahmann

The lack of reliable, nondestructive part qualification for additively manufactured (AM) parts hinders their adoption in key industries of national interest, such as aerospace and defense. Resonant ultrasound spectroscopy (RUS) is a relatively low-cost and nondestructive method for accurately determining material properties. Here we explore potential applications for using RUS to derive the material properties of AM parts as an alternative to computationally expensive inversion, as well as attempt to identify the quality and different printing settings of these parts post-processing. We perform RUS on 121 cylinders manufactured via laser powder bed fusion (LPBF) from Aeromet’s A20X alloy. We find the modal frequencies of these cylinders, first via unconstrained finite element (FE) models in ABAQUS, then through a minimally constrained experimental setup. In our FE models, we vary density, elastic parameters, and symmetry of cylinders in anticipation of variations in the LPBF build process of our physical cylinders. We attempt to cluster the simulated eigenfrequency data from these cylinders. Clusters are then used to train a supervised model for experimental spectral data classification. Drawing on the relationship between modal frequencies and directional stiffness, we propose that our model’s assignment of printed cylinders into clusters should allow us to relate their material properties to the known properties of simulated cluster members, with accuracy as further tested and discussed. This model will attempt to skip the costly inversion process and directly calculate sample parameters in a computationally efficient way, compared to the standard inversion process. Our results inform how practical and reliable the union of RUS and FE spectra are for material property derivation and part qualification of simple AM parts.

Increasing Multi-Axis Testing Confidence through Finite Element and Input Control Testing

Mentors: Sandra Zimmerman, John Schultze, Peter Fickenwirth
Students: Kaelyn Fenstermacher, Sarah Johnson, Aleck Tilbrook

Testing devices multi-axially is a better approximation of a device’s operation conditions compared to single-axis testing. However, multi-axis environmental tests involve a more complicated test set up that is often determined by a test engineer’s judgment. Some of the complexities the engineer must determine include where to locate shakers on complex geometries, how much input in required to excite the structure, how many control channels are necessary, and how well the environmental boundary conditions replicate the operational environment. Any one of these things can lead to the device under test (DUT) response differing greatly from the desired response. Often there is no indication of whether the environmental test set up appropriately replicates field conditions prior to the start of the test. The ability to simulate a test, for example on a validated finite element model (FEM), would allow the engineer to have insight into the predicted response of the DUT prior to performing an environmental test. This study uses Sandia National Laboratories Rattlesnake control software to conduct virtual tests on an FEM of the base section from a Box Assembly with Removeable Component (BARC) and aims to determine if virtual testing can be used to predict optimal environmental test set up accurately and achieve the desired response from the DUT. This method could provide test engineers with a reliable pretest standard to determine input controls and locations, equipment requirements, and sensor placement all prior to performing a test, reducing the time required in the lab and improving test results.

Neural Radiance Fields as Digital Twins for Photorealistic Event-Based Structural Dynamics

Mentors: Alessandro Cattaneo, Moises Felipe Mello da Silva, Fernando Moreu (University of New Mexico)
Students: Marcus Chen, Allison Davis, Edward Walker

Digital twins are virtual representations of real-world structures that can be used for modeling and simulation. Because of digital twins’ ability to simulate complex structural behaviors, they also have potential for structural health monitoring (SHM) applications. Video-based SHM techniques are advantageous due to the lower installation/maintenance costs, analysis in high-spatial resolution and its non-contact monitoring features. Both digital twins and video-based techniques hold particular interest in the fields of non-destructive evaluation, damage identification, and modal analysis. An effective use of these techniques for SHM applications still poses several challenges. Neural Radiance Fields (NeRFs) are an emerging and promising type of neural network that can render photorealistic novel views of a complex scene using a sparse data set of 2D images. Originally, NeRF was designed to capture static scenes, but recent work has extended its capability to capture dynamic scenes which has implications for medium and long-term SHM. However, to-date most NeRFs use frame-based images and videos as input data. Frame-based video monitoring approaches result in redundant information derived from the fact that, for structural dynamics monitoring, only a small number of active pixels record the actual dynamical changes in the structure, resulting in intensive computational loads for data processing and storage. A promising alternative is event-based imaging, which only records pixel-wise changes on the illumination of a scene. Event-based imaging creates a sparse set of data, while accurately capturing the dynamics. The work proposes a method to extract the dynamics of a structure through simulated loads on a digital twin. The simulated model generated using NeRF methods enabled a fully editable digital twin that could simulate different loads. The digital twin model was then used along with an event camera simulator to generate event-based data. The authors extracted modal information of the structure with manifold learning and blind source separation methods. Validation was performed on a structure of known dynamics using event-based cameras.

Structural Health Monitoring in the Context of Non-Equilibrium Phase Transitions

Mentors: Francesco Caravelli, Chuck Farrar, Cynthia Reichhardt
Students: Luis Corral Cano, Juan Molinar, Joseph Sommer

A statistical pattern recognition paradigm for a structural health monitoring (SHM) system can be defined as: operational evaluation, data acquisition, feature selection and extraction, and statistical classification. Currently, feature selection is primarily based on physics-informed engineering judgement. The goal of this work is to develop a more principled approach for feature selection. This study uses the concept of increased complexity of a system that results from structural degradation and identifies this change in complexity using coherence, cross-correlations, Shannon entropy and residual errors from autoregressive models. The unique aspect of this study is its demonstration that these features are consistent with recent developments in statistical physics associated with fluctuation dynamics and non-equilibrium (NE) phase transformations (PT). In the context of SHM, a NEPT consists of an irreversible process (e. g. crack formation, yielding) characterized by an energy-absorbing transformation process. Recently NE fluctuation theorems define expected behaviors of systems exhibiting NEPT related to entropy production, work and ground state energy, scaling characterization of intermittent NE fluctuations, and extreme statistics of these fluctuations. The hypothesis of this study is that the features defined above are consistent with NE fluctuation theorems and, hence, these theorems provide a more principled approach to defining damage-sensitive features. The theorems state that if there is a NEPT (damage) and associated energy dissipation, then there are corresponding changes in fluctuation response. To calculate entropy production and create a link to recent studies of NE fluctuation theorems, experimental and numerical tests are conducted on systems with NEPTs. Furthermore, based on other studies of NE fluctuations, the statistical physics community has shown that a variety of physical (e. g. biological) and non-physical (e. g. economic) systems follow common trends leading to universal properties at the NEPTs and we investigate these properties to see if they can guide required sensor locations and define the time scales associated with system transitions to damaged conditions.

2021

Projects

Simulating SDOF and MIMO Vibration Tests

Mentors: Scott Ouellette, Thomas Roberts
Students: Sebastian Chirinos, Aneesh Pawar, Haley Tholen

Single-axis vibration testing is the current industry standard for structural system qualification. While this method is well studied and documented, there are several shortcomings that are now being addressed using Multiple Input/Multiple Output (MIMO) testing strategies. In this project, analytical and FEA models of a BARC single-axis vibration test will be developed to: 1) validate a broadband dynamics model of the BARC system by parametrically studying various methods for applying an input forcing spectrum, and 2) exploring untested BARC configurations and input locations for simulating various service environments.

Environmental-Insensitive Damage Features Based on Output-to-Output Coherence

Mentors: Chuck Farrar, Garrison Flynn
Students: Samuel Hinerman, Makalya Ley, Peter Newman

One of the biggest challenges that prevents structural health monitoring research from transitioning to practice is that in situ operational and environmental (O&E) variability can produce changes in the measured system response that can be mistaken for damage. Therefore, a successful SHM system must be robust to changing O&E conditions while maintaining high sensitivity to damage. During LADSS 2020, we began approaching this challenge evaluated causality-based features (Granger causality, transfer entropy, and coherence) to test their ability to identify damage while remaining insensitive to global O&E variability (Gibbs et al. 2021). These metrics showed some promise for two experimental and one simulated data set. Coherence was found to have the most consistent results, showing potential for both damage detection and O&E insensitivity. Mahalanobis distance of the coherence vectors was able to quantify changes to the coherence as damage was introduced in the form of a crack, both with and without O&E variability. However, poor signal-to-noise ratio results in decreased coherence, and as a result, decreased sensitivity to damage. In this project, the team will pursue a comprehensive study of damage features leveraging output-to-output coherence and validate its ability to identify damage in the presence of O&E variability.

Signature-Driven Adaptable Collaborative Sampling

Mentors: Jeff Tippmann, Christian Ward, Brian West
Students: Erik Barbosa, Sean Detwiler, Kristen Steudel

Structural health monitoring (SHM) commonly involves recording vast amounts of data from multiple sensors. Depending on the specific application these systems can require prohibitively large bandwidth for resolving damage-indicative features. The primary challenges are often attributed to memory limitations, energy requirements, cost, weight, and size. Previous research in the Engineering Institute has focused on compressive sensing to reduce the size of data transferred while maintaining signal reconstruction capabilities for high-fidelity signal analysis. Furthermore, there has been recent work in understanding event-based monitoring, which collects data non-uniformly based on triggers. Both modalities have aimed to understand how to improve sampling for multi-sensor systems, but are restricted to lower sampling rates or are prone to missed events. This project will focus on determining optimal triggering schemes of finite length recorded signals in order to expand the bandwidth beyond Nyquist, assuming the signal of interest is omnipresent across all recordings. Here we will explore the deliberate spacing of triggered uniform sampling rate data collection from single and multiple channels over different modalities (e.g. acoustic, vibration, and/or current). The students will demonstrate the developed approach through an experiment designed by the students and executed by the mentors in the lab.

Estimation of Light Intensity Time Series from Event-Driven Imagery for Acoustic Measurements

Mentors: David Mascareñas, Alessandro Cattaneo, Fernando Moreu
Students: Celeste DeVilliers, Jack Sorensen, Kevin Zheng

Event-driven imagers are attractive for structural dynamics because they are low-power, high information bandwidth, and high-dynamic range. Current event-driven imagers are capable of detecting the light intensity changes associated with an LED blinking at 18 kHz. This implies that event-driven imagers are capable of capturing dynamic motion associated with the human range of human hearing. The trajectory of the technology is such that it is reasonable to expect this bandwidth will continue to increase as time goes on. As a result, event-driven imagers are a very attractive potential replacement for high-speed imagers for structural dynamics applications. The primary issue with using event-based imagers for structural dynamics applications in that currently the majority of statistical techniques developed for structural dynamics assume data is captured with a uniform sampling rate. In order to enable widespread use of event-driven sensors for structural dynamics applications it will be necessary to develop an acceptable technique for converting event data to time series which can be sampled uniformly. An additional issue with event-driven imagers is that the event generation behavior between pixels on the same imager can exhibit significant variations. These variations need to be better understood and accounted for. Background research has found that issues related to both of these needs have been addressed by the neuroscience community. In this project we will consider the spike-based data analysis techniques that have been developed in the neuroscience community and investigate their applicability to structural dynamics in the context of event-driven imagery. We will then use these techniques as inspiration to develop event-based data processing techniques to tie event-based data to existing structural dynamics analysis frameworks.

Waveometry: Multi-Layer Input Deep Learning Applied to Ultrasonic Wavefield Measurements

Mentors: Adam Wachtor, Erica Jacobson, Nikolaos Dervilis
Students: Justin Dalton, Nicholas, Dzomba, Cole Maxwell

Full-field nondestructive evaluation (NDE) technologies provide a holistic assessment of the damage state of a part or assembly. Since damage can occur anywhere from the factory floor (improper layup in a wind turbine blade) to its service environment (material thinning caused by corrosive material storage or transport), such NDE is needed to certify a part or assembly associated with high-risk applications both before and after it is placed into service to ensure its current state meets the specifications it was designed to. Most of these techniques rely on repetitious transient measurements at a moving point to build up the full-field assessment – equating to long and costly measurement times which motivates many industries to abandon full-field NDE and rely on point inspections at discrete locations instead.

Acoustic steady-state excitation spatial spectroscopy (ASSESS) is a LANL-developed, rapid NDE technology that is positioned to address this issue for a wide variety of industries and LANL applications. ASSESS uses steady-state ultrasonic excitation and noncontact laser Doppler vibrometer (LDV) measurements of the surface response to identify damage orders of magnitude faster than traditional time-of-flight ultrasonic measurements. ASSESS measurement speeds, 5-10 m2/min, enable full wavefield images to be collected over large parts and assemblies.

A convolutional neural network (CNN) is a deep learning algorithm which can take in input data e.g. an image most famously, allocate importance to various aspects/objects in the image and then be able to differentiate/classify one from the other. CNN processing of the wavefield images generated from ASSESS measurements has several advantages over other processing methods for identifying and characterizing defects, including processing speed and defect predictions nearpart boundaries.

Deep Reinforcement Learning for Active Structure Stabilization

Mentors: Alex Scheinker, Alan Williams
Students: Will Compton, Mason Curtin, Wilson Vogt

Deep reinforcement learning (RL) is an active area of machine learning in which optimal control methods are applied to analytically unknown systems by modeling the system dynamics or optimal control policies using deep neural networks. Deep RL methods have shown extremely impressive results for systems with fixed rules, such as chess. Our system of focus will be a 4-story structure attached to a shake table as well as several smaller active controllers. The goal is to develop deep RL-based active feedback control tools, to apply forces dynamically to the floors in the structure, which can mitigate disturbances like earthquakes simulated by the shake table on the ground floor.

Estimation of Structural Vibration Modal Properties Using a Spike-Based Computing Paradigm

Mentors: David Mascareñas, Andrew Sornberg
Students: Jabari Allen, Raymond Chu, Troy Sims

The goal of this project is to develop a spike-based computational pipeline for unsupervised learning of vibration mode shapes. This is of interest because emerging spike-based computing resources consume significantly less power than conventional digital electronics. This is important for designing next-generation smart structures as well as persistent structural health monitoring. It is possible to extract structural modal properties from data generated by a structural simulation by solving a blind source separation unsupervised learning problem using a conventional digital computer. One of the primary goals of this work is to be able to automatically extract the natural frequencies, mode shapes and damping ratios of a structure when exposed to dynamic loading in a low-power manner using spike-based computing. By focusing on structural systems the team will immediately be able to create models of the structure to simulate dynamic response and the associated sensor voltage signals that would be processed downstream by a spike-based learning algorithm. The structural simulations will allow the generation of simulated sensor voltage signals. These signals can then be delivered to a spike-based computing simulator to develop/train and test the spike-based unsupervised learning algorithm to solve the structural dynamic system identification problem. Use of the simulation will facilitate the consideration of a wide variety of loading conditions during development and training. The team will take inspiration from the computational neuroscience research community as they develop their algorithm. The results of this work will applicable to energy infrastructure, pressure vessels, aerospace, prosthetics and civil infrastructure.

Data challenges for Structural Health Monitoring of Electrical Machines

Mentors: Nick Lieven, Phil Cornwell
Students: Alex Binder, Conner, Ozatalar, Kendyl Wright

All industrial plant deploy rotating machines to convert electrical input to reactive loads. Necessarily for large electrical machines the requirement is to use three-phase power and usually operate at a synchronous rate coupled with the power network frequency (60 Hz in U.S., 50 Hz in U.K.). This operating environment presents a significant challenge to structural health monitoring, namely the overwhelming influence of the power-grid frequency on the electrical signal measured in comparison to the structural and electrical performance of the machine. The project will explore both the nature of the dominant power-grid frequency, the slip frequency associated with rotation of the armature and the structural dynamics response including the 1/rev rotation speed of the machine’s shaft. The objective is to remove the obscuring closely coupled power-grid “carrier” signature and associated signal processing artifacts in an effort to observe the true dynamic response signature of the machine.

Lectures

Effective Presentations

Phil Cornwell
Professor of Mechanical Engineering
US Air Force Academy

In this talk, Dr. Cornwell will discuss how to give effective oral presentations using a type of slide design called the assertion-evidence approach. In this approach, the presenter is required to explicitly identify what messages (assertions) he or she is trying to make and the evidence to support these assertions. The information in this talk is based on the work of Dr. Michael Alley from Penn State.

B61-12 Life Extension Program: System Qualification and Joint Testing

Curtt Ammerman
LANL, Q-15

Los Alamos National Laboratory is the nuclear design agency responsible for the B61-12 Life Extension Program (LEP). The term “life extension program” means a program to repair/replace components of a nuclear weapon to ensure its ability to meet military requirements. By extending the life, or time that a weapon can safely and reliably remain in the stockpile without having to be replaced or removed, the National Nuclear Security Administration (NNSA) is able to maintain a credible nuclear deterrent without producing new weapons or conducting new underground nuclear tests. This LEP will consolidate multiple B61 mods, replace aging components, and extend the lifetime of the B61 for an additional 20 to 30 years. The LEP will complete a first production unit no later than the end of FY2020. This presentation will provide a brief overview of the B61LEP, and will dive more deeply into the testing and qualification activities that are being conducted to ensure that the B61-12 meets its requirements to be safe and reliable.

An Overview of Offshore Wind Energy

Nikolaos Dervilis
Dynamics Research Group, Department of Mechanical Engineering
University of Sheffield

The use of offshore wind farms has been growing in recent years, especially in Europe. United Kingdom is presenting a geometrically-growing interest in exploring and investing in such offshore power plants as the country's water sites offer impressive wind conditions. The new generation of offshore wind turbines shall have blades that will exceed 100m and soon will reach 150m in size (15MW wind turbine PM generators). This talk will give an overview of offshore wind turbines and farms.

The cost of an offshore wind farm is relatively high, and therefore their reliability is crucial if they ever need to be fully integrated into the energy arena.

In turn, this talk will present an overview of current monitoring trends for wind turbines (WTs) and will try to address the motivation and the effectiveness of Structural Health Monitoring (SHM) and machine learning applications for WTs, as well as, the idea of intelligent WT.

Ratchet Effects and Non-Equilibrium Dynamics

Cynthia Reichhardt
LANL, T-1

I give an introduction to the topic of ratcheting motion, a fundamentally nonequilibrium phenomenon. In a ratchet effect, a net dc transport of particles, dislocations, or other objects is produced by purely ac driving in the presence or absence of thermal fluctuations. I describe the crucial role that symmetry breaking plays in producing ratcheting motion, and show examples drawn from fields as diverse as swimming bacteria and superconducting magnets, and including the thermal ratcheting of dislocations in materials. For collectively interacting systems, the direction of dc motion can show multiple reversals.

An Introduction to Robotic Motion Planning

Beth Boardman
LANL, E-3

This presentation will introduce robot motion planning concepts, including discrete and continuous motion planning algorithms. Briefly, topics in dynamic motion planning will be covered. The presentation will conclude with some examples of robotic projects at the Laboratory.

Aircraft Structural Dynamics

Nick Lieven
University of Bristol

Prof Lieven’s talk will explore the extreme dynamic behavior of aircraft and their related systems. A particular focus will be on the physical factors which lead to flutter and aircraft instability. Although flutter can be an entirely predictable phenomenon there is an increasing awareness that such aeroelastic instabilities can be caused by structural non-linearities and human intervention. The talk will explain how these factors can interact in a potentially destructive way and what technologies and design modifications can be deployed to mitigate against this behavior. The talk will consider both the modelling aspects of aircraft non-linearities and the practical considerations associated with flight testing and – ultimately – flight safety.

Experimental Modal Overview

Pete Avitabile
University of Massachusetts Lowell

The objective of the experimental modal overview following the dynamic systems lecture is to demonstrate through a live impact modal test to identify a few important theory items, discuss impact test considerations, acquire a set of FRFs and obtain modal parameters. Discussion of the process as outlined above to conclude the experimentation/demonstration.

2020

Projects

Deep Waves: Defect Characterization in Ultrasonic Wavefield Measurements using Deep Learning

Mentors: Adam Wachtor, Erica Jacobson
Students: Josh Eckels, Isabel Fernandez, Kelly Ho

For many industries, nondestructive evaluation (NDE) is essential to ensuring the safety and performance of the part or system (e.g. equipment, vehicle, or infrastructure) they manufacture or maintain. NDE is often utilized before a part or system is ever placed into service to ensure its manufacturing meets the specifications it was designed to, but is also used after the part or system has been placed in its service environment to ensure it should remain there or be pulled for repair. Downtime once a part or system has been placed into service is very costly, so minimizing the time for NDE to determine if it still meets requirements is very advantageous. Acoustic wavenumber spectroscopy (AWS) is a rapid NDE technique that utilizes steady-state ultrasonic excitation and noncontact laser Doppler vibrometry measurements of the surface response to identify damage orders of magnitude faster than traditional time-of-flight ultrasonic measurements [1, 2]. The speed of this technique allows for full-field wave maps of the part/system to be collected. This project will leverage deep learning and convolutional neural network (CNN) techniques, often applied for image analysis and recognition, to identify and characterize damage from simulated AWS measurements.

Designing Operations- and Environmental-Insensitive Damage Features

Mentors: Chuck Farrar, Garrison Flynn
Students: David Gibbs, Kaleb Jankowski, Ben Rees

Detecting Acoustic Change in a Reverberant Environment

Mentors: Jeff Tippmann, Brian West
Students: Michael Gassen, Mitchell Roberts, Ian Marts

The detection of a wave propagation perturbation in a medium has been of interest to a large range of applications. Often times the perturbation is an addition of a scatterer the medium, such as a crack in the structure during an ultrasonic measurement, SONAR detection of an underwater vehicle, or equipment in manufacturing or plant setting utilizing a condition-based monitoring system. However, if a problem existed where the removal or movement of an object is of interest, then this problem can be significantly more challenging particularly in a medium with many existing sources or other objects. It has been shown a perturbation can be detected in a random forest with no prior knowledge of the medium. In this project, we will aim to extend these results but to a reverberant environment and develop techniques for employing arrays of microphones for such detection in a real-world setting.

Digital Spatio-Temporal Coded Exposures for Dynamic Characterization using Event-Driven Imagery

Mentors: David Mascareñas, Alessandro Cattaneo
Students: Andrew Gothard, Corey Henry, Daniel Jones

In recent years the use of computer vision techniques for structural health monitoring has become of increasing interest. One particular issue that must be considered when making a measurement of a dynamic system using an imager is the effect of the frame-rate, and shutter speed on the phenomena of motion blur in the image. The field of computational photography has introduced the concept of the “coded exposure” to address some of the issues associated with motion blur. However, these techniques generally require adding specialized hardware to imagers in the form of fast, programmable shutters. In recent years, biologically-inspired, event-based neuromorphic imagers have become increasingly popular. Neuromorphic imagers present an interesting opportunity for implementing coded exposures in a purely digital manner in software. Furthermore, these digital coded exposures can be applied on a pixel-by-pixel basis so that different spatial sections of an image can focus in on dynamics at different time-scales. Furthermore, since these coded exposures are digital they can be made to adapt as the time scales present in the individual pixels also change. The goal of this work is going to be to study the applicability of using temporal and spatio-temporal coded exposures for decomposing motion at different time and length scales in video. We will start exploring what spatio-temporal resolution test targets should look like for dynamic imagers. Conventional imagers require static resolution test targets but we hypothesize that dynamic imagers should inherently have spatio-temporal test targets whose design is informed by structural dynamics theory. This problem is applicable to the extraction of mode shapes and natural frequencies from imager data taken from vibrating structures.

Hidden Layers: Detection of Composite-Metal Interface Delamination using Ultrasonic Wavefield Imaging

Mentors: Eric Flynn, Ian Cummings
Students: Casey Gardner, Michael Koutoumbas, Young Ko

The material needs for the aerospace industry have driven the development of composite metal laminates (CMLs) for a wide range of applications. CMLs contain bonded layers of composites and metals, and are often utilized because they offer weight savings while exhibiting high strength and corrosion resistant properties. These materials present interesting inspection challenges due to the heterogeneous nature of the material. Of specific interest is the capability to detect delamination between the composite and metal layers from the composite side of the material. In fact, composite-metal delaminations in carbon-fiber wrapped aluminum fuel tanks were the suspected cause of the explosive destruction of SpaceX’s Falcon 9 rocket on September 1, 2016. This project will apply simulation and data analysis techniques to ultrasonic inspection measurements via an acoustic wave spectroscopy (AWS) system to a CML sample and involve the development of signal processing techniques to identify areas of composite-metal delamination.

Informed Design of Radiation Detecting Textiles

Mentors: Margaret Root, Sarah Sarnoski
Students: Ellie Andreyka, Nai'a North, Anand Iyer

The objective of this research is to study the effects of radiation on optically stimulated luminescence (OSL) fibers, which can be used for mobile radiation detection, with the ultimate goal of applying them to textiles with real-time radiation detection functionality. Wearable clothing with radiation detection that is fast and accurate would revolutionize the field of radiation protection and open a major new technology development focus area. However, such advanced functionality will require significant engineering design and testing to create highly complex, integrated products for practical use in working environments. Successful outcomes for this proposed work will be simulation-based determination of the signal processing and radiation detection characteristics of the optical fibers.

Robust, Self-Healing Video-Based Sensor Networks

Mentors: Alessandro Cattaneo, David Mascareñas
Students: Sarah Mantell, Addison Schwamb, Nathan Sapong

Our goal is to develop robust, flexible, video-based structural dynamics identification algorithms/techniques which will usher in a new paradigm for pervasive structural integrity monitoring of civil infrastructure. The cost per spatial data point associated with using a network of imagers to monitor infrastructure integrity is multiple orders of magnitude less than the costs associated with using traditional contact sensors. As a result, it is relatively trivial to add additional imagers to endow the sensing network with redundancy in case an imager is lost. We will perform research on the development of structural health monitoring algorithms that can self-heal in the event an imager is lost or the imager position is changed. We will perform research to quantify the impacts of these losses in order to choose imager locations that result in a robust structural health monitoring system. It is likely that during the multi-decade lifetime of an infrastructure sensing network that a camera mount might get bent slightly out of place. To mitigate the effect of these types of perturbations we will perform research on self-healing sensing network data processing techniques to detect the disturbance to the measurement process, and re-register data so that imagery from the sensor can still be used for structural health monitoring analysis.

Lectures

Making Effective Presentations - Rethinking Slide Design

Phil Cornwell
Professor of Mechanical Engineering
Rose Hulman Institute of Technology

B61-12 Life Extension Program System Qualification and Joint Testing

Curtt Ammerman
LANL, Q-15

An Overview of Offshore Wind Energy

Nikolaos Dervilis
Dynamics Research Group, Department of Mechanical Engineering
University of Sheffield

The cost of an offshore wind farm is relatively high, and therefore their reliability is crucial if they ever need to be fully integrated into the energy arena.

Verification and Validation of Computational Models

François Hemez
Lawrence Livermore National Laboratory

Verification and Validation (V&V) refers to a broad range of activities that are carried out to provide evidence that measurements and predictions are credible and scientifically defendable. These lectures introduce the main concepts of V&V and illustrate how various V&V activities can be carried out for engineering applications. V&V activities include software quality assurance, code and calculation verification, data analysis and archiving, sensitivity analysis, model calibration, and the quantification of uncertainty. The cornerstone of V&V is threefold with, first, showing whenever possible that numerical predictions are accurate relative to physical data over a range of settings or operating conditions; second, quantifying the sources and levels of uncertainty; and, third, demonstrating that predictions are robust, that is, insensitive, to the modeling assumptions and lack-of-knowledge. Examples are presented in solid mechanics, transient response of structures, and shock physics.

Effective Microstructure on Damage and Failure in Materials

Saryu Fensin
LANL, MST-8

For ductile metals, the process of dynamic fracture occurs through nucleation, growth and coalescence of voids. For high purity single-phase metals, it has been observed by numerous investigators that voids tend to heterogeneously nucleate at grain boundaries and all grain boundaries are not equally susceptible to void nucleation. However, the reasons behind this observation is not fully understood. It is reasonable to assume though that grain boundary structure and its affect on related properties must play a key role in understanding this deterministic relationship. In this work we explore grain boundaries properties like energy and excess volume, in addition to its interactions with dislocations and investigate any relationship it might have with dynamic fracture. We will attempt to compare the mechanisms behind void nucleation in FCC (Cu) and BCC (Ta) materials by using molecular-dynamics simulations.

Aircraft Structural Dynamics

Nick Lieven
University of Bristol

LANL Satellite Engineering

Amy Regan
LANL, ISR-5

LANL satellite engineering comprises a broad set of disciplines working together to bring science ideas to fruition. ISR Division’s mission is to create, deliver, support and exploit innovative sensing systems for space-based, airborne and ground-based applications to address critical national security and scientific challenges. LANL has delivered more than 1400+ sensors on 400+ instruments on 60+ satellites. This talk will discuss the various engineering challenges to putting hardware in space.

2019

Projects

Applying Information Complexity Measures to Structural Health Monitoring

Mentors: Chuck Farrar, Alex Scheinker, Nick Lieven
Students: Greg Mellos, Salma Leyasi, Hannah Donajkowski

The process of implementing a damage detection strategy for aerospace, civil and mechanical engineering infrastructure is referred to as structural health monitoring (SHM). The SHM process compliments traditional nondestructive evaluation by extending these concepts to online, in-situ system monitoring on a more global scale. SHM can best be described in terms of a statistical pattern recognition paradigm. In this paradigm, the SHM process can be broken down into four parts: (1) operational evaluation, (2) data acquisition and cleansing, (3) feature selection and extraction, and (4) statistical model development for feature discrimination. Studies to date suggest that a fundamental axiom of SHM is that all damage increases the “complexity” of a system. This increase in complexity can manifest itself in terms of geometric complexity, material complexity, or information complexity encoded is sensors monitoring the structure’s dynamic response. The challenge is to determine what are the appropriate measures of complexity to be used for a given damage detection problem. This project will focus on studying the various measures of information complexity based on the concept of “entropy”. This study will begin by studying the Shannon entropy, which was developed for communications theory in the late 1940s. Since then a number of information entropy measures have been proposed in the literature for a variety of applications. They include: Komogorov-Sinai entropy, Pesin formula, permutation entropy, Renyi entropy, topological entropy, transfer entropy, spectral entropy, differential entropy, conditional entropy, relative entropy, and mutual information.

Assessing Predictive Capabilities for Nonlinear Dynamic Structural Responses

Mentors: Kyle Brindley, Andy Morello, Thomas Allard
Students: Brian Evan Saunders, Liliana Haus, Jonathan Acosta

Predictions of dynamic structural responses of complex assemblies subject to engineering environments are commonly made with simulations of finite element models. The assemblies and their models often contain material, interfacial, or geometric features that produce nonlinear responses that pose challenges to finite element analysis tools. A previous study experimentally investigated the linear and nonlinear dynamic responses of a four degree-of-freedom (DOF) benchmark testbed configured for variable states of intermittent contact. This study developed a corresponding numerical, lumped-mass model and performed associated verification and validation1. The objective of this project is to extend this past work in order to develop a finite element model for the testbed, explore effects of available contact algorithms/parameters, and to identify sources of uncertainty arising from finite element techniques.

Characterizing the Electrical Pathways of Heart Cells from Calcium Imaging Video – An Unsupervised Machine Learning Approach

Mentors: Bridget Martinez, David Mascareñas, Pulak Nath, Jennifer Harris, Kent Coombs
Students: Lauren Schneider, William Anderson, Richard Yeong

This project will focus on the development of multi-modal, high-resolution unsupervised machine learning approach applied to video of calcium transport imaging. The purpose of which is to study the electro-mechanical phenomena of cardiac automaticity which allows for individual motions of heart cells (cardiac myocytes) to generate proper cardiac contraction, which is automatic and synchronized. The results of this study will find its utility in potentially differentiating between cardiac myocytes that are in electro-mechanical synchrony and those which are not – a useful assessment when considering accessory pathways and subsequent pathology (described below). Full-field, high-resolution, anomaly detection methods for non- contact, full-field electro-mechanical dynamics of cardiac myocytes currently do not exist. This work will be enabled by recent advances in the LANL Engineering Institute in unsupervised machine learning algorithms for extracting high resolution dynamic mode shapes from video of vibrating structures. It also leverages LANL’s artificial organ program - Advanced Tissue- engineered Human External Network Analyzer (ATHENA) team work on calcium imaging of cardiac myocytes.

Power Sync

Mentors: Eric Flynn, Adam Wachtor, Nick Lieven
Students: Jessica Chan, Ferrill Rushton, Eugene Lin

Phase-sensitive vibration measurements, such as those used for modal analysis, wave propagation studies, and structural health monitoring, require precise time synchronization between sensors. However, time synchronization across multiple, spatially distributed data acquisition units in GPS- and/or radio-denied environments is an ongoing challenge.

The US power grid is divided into three interconnections: Eastern, Western, and Texas. Throughout each interconnection, the voltage waveform is nearly identical, operating at a nominal 60 Hz. However, the actual frequency varies by tens of millihertz over the course of a minute and up to hundreds of millihertz over the course of day. Hypothesis: The pattern of variation in power grid frequency, which should be identically observed at any measurement point within an interconnection, can serve as a common reference signal for time-synchronizing distant vibration measurement systems.

Real-Time Hybrid Substructuring for System Vibration Testing

Mentors: Garrison Flynn, John Schultze, Stuart Taylor
Students: Manuel Serrano, Tim Shenouda, Nadim Bari

Vibration tests in a controlled laboratory setting are typically used to qualify systems for deployment in their real-life environments. System complexity often leads to increased difficulty qualifying specific components. Real-time hybrid substructuring (RTHS) is a methodology for interfacing physical experiments with numerical simulations in real-time to improve the accuracy of dynamic testing. RTHS is traditionally used in early stage design of systems with the physical substructure being a difficult to model component and the remainder of the system being the numerical substructure. This project will use RTHS in the opposite fashion to better inform component qualification tests. In this project, the main system will serve as the physical substructure and the component of interest will be numerically modeled. This approach allows for effects of uncertainty due to jointed connections to be represented numerically by probabilistic stiffness and damping coefficients in the numerical substructure. The numerical substructure will be a predictive tool to determine if a component is reaching or exceeding a desired threshold, with quantifiable uncertainty, thus providing important feedback to vibration inputs of the physical substructure.

Toward Extracting Multidimensional Kinematics from SWIFT Experiments

Mentors: Christopher Tilger, Michael Murphy, Michael Bowden
Students: Ethan Billingsley, Yolnan Chen, Robert Billette

The shock wave image framing technique (SWIFT) developed at LANL is a versatile flow- visualization diagnostic that allows real-time ultra-high-speed capture of explosive-driven shock waves in transparent witness media. Recent advances in data-analysis techniques for detonator and bare-explosive loading on poly(methyl methacrylate) (PMMA) samples have shown that unique one-dimensional shock-wave trajectories can be defined and used to obtain accurate shock kinematics along the charge centerline. An extension of the analysis to off-centerline regions is needed to attempt performance characterization across the working surfaces of explosive components. This need offers a unique inverse-design problem that favors creative solutions within the physics-based constraints of compressible fluid mechanics, similarity analysis, and shock physics.

Visio-Acoustic Data Fusion

Mentors: Alessandro Cattaneo, Jeffery Tippman, David Mascareñas
Students: Chad Samuelson, Christopher Whitworth, Caitrin Duffy-Deno

This project focuses on the development of visio-acoustic data fusion techniques to bring together the complementary advantages of acoustics and video-based structural dynamics to enable new structural characterization techniques. Acoustic arrays have found great utility for remotely localizing radiators of sounds generated from faulty equipment. One of the major challenges associated with acoustic techniques is that their ability to discriminate between sources closely spaced together is limited by the wavelength of the sound in air. In the last few years high- resolution, full-field structural identification techniques from video of vibrating structures have been developed. These techniques have demonstrated very high spatial-discrimination capabilities. The challenge with using imagers to capture acoustic radiators is that acoustic signals generally occur in the kHz range which in many cases requires the use of powerful illumination of a scene in order to receive enough photons at the imager to capture structural dynamics.

The LANL Engineering Institute has shown that full-field, high resolution mode shapes can be captured using imagers sampling both sub-Nyquist as well as in a compressive sampling mode. We speculate that it may be possible to use measurements from an acoustic array to inform an adaptive sampling scheme for an imager so that relatively low-speed cameras can be used to perform high-spatial resolution discrimination of acoustic radiators. The focus of this project is the development of the data fusion techniques for combining measurements from acoustic arrays and imagers. To date imagers are only used in conjunction with acoustic arrays for visualization purposes. In this work we aim to intimately couple high-frequency acoustic information with high- resolution imagers in a complementary fashion to mitigate the shortcomings associated with each. Visio-acoustics could enable new capabilities for characterizing the health of machinery and structures.

Lectures

Making Effective Presentations - Rethinking Slide Design

Phil Cornwell
Professor of Mechanical Engineering
Rose Hulman Institute of Technology

B61-12 Life Extension Program System Qualification and Joint Testing

Curtt Ammerman
LANL, Q-15

An Overview of Offshore Wind Energy

Nikolaos Dervilis
Dynamics Research Group, Department of Mechanical Engineering
University of Sheffield

The cost of an offshore wind farm is relatively high, and therefore their reliability is crucial if they ever need to be fully integrated into the energy arena.

Verification and Validation of Computational Models

François Hemez
Lawrence Livermore National Laboratory

Effective Microstructure on Damage and Failure in Materials

Saryu Fensin
LANL, MST-8

Aircraft Structural Dynamics

Nick Lieven
University of Bristol

LANL Satellite Engineering

Amy Regan
LANL, ISR-5

2018

Projects

Imager-based Characterization of Viscoelastic Material Properties

Mentors: Jason Anderson, Chuck Farrar, Bridget Martinez, Yongchao Yang (Remote), David Mascareñas
Students: Tia Kauppila, Kayla Wielgus, Howard Brand

The Laboratory mission requires accurately modeling the behavior of components containing energetic materials (i.e. explosives). Unfortunately, characterizing the behavior of energetic materials is complicated by safety requirements that stipulate that only a maximum size of material may be used during testing. The constraint on maximum size of material that can be tested can significantly impact the ability to characterize important material properties using conventional techniques. In this work we will develop imager-based techniques to extract modal damping information that can be used to infer the viscoelastic properties of materials such as explosives. These techniques will be based on video processing advances made at the Lab's Engineering Institute that allow natural frequencies, mode shapes, and damping ratios to be extracted from video of a structure vibrating. The team will build an experimental setup to measure the viscoelastic properties of cubes of rubber-like materials with dimensions on the order of a few centimeters. We anticipate that a mass will be placed on top of the rubber material. A tapping device will also be included in the experimental setup to provide a repeatable excitation to the rubber-mass structure. The team will create an experimental imager setup to observe the rubber-mass structure. The team will also create a model of the rubber mass structure. The team will start their experiments using larger rubber samples so the results obtained with the imaging technique can be compared with the results from conventional techniques. The ultimate goal is for the team to demonstrate their technique on a length scale that is appropriate for explosive testing.

Imager-based Techniques for Analyzing Metallic Melt Pools for Additive Manufacturing

Mentors: David Mascareñas, Thomas Lienert, Bridget Martinez, Garett Kenyon, Yongchao Yang (remote)
Students: Cedric Hayes, Caleb Schelle, Gregory Taylor

The goal of this work is to develop an online, imager-based qualification and control system for manufacturing processes that use a melt pool (GTA/laser welding, additive manufacturing). This work will be based on Lab-developed algorithms for automatically characterizing the dynamics of vibrating structures at high spatial resolution using video. Significant research on using high-speed video to characterize melt pools has been done in the past, but until now there has never been a systematic technique for extracting structural dynamics information from the video that can be used to infer otherwise indiscernible geometric properties of melt pools that can be used to quantify manufacturing quality and control the process for improved performance. Furthermore, the advent of event-based imagers opens the door to enabling real-time qualification and control of additive manufacturing processes. In this work the team will use conventional imagers, event-based imager and perhaps also thermal imagers, in combination with other sensors to characterize the physical and geometric properties of a pool of liquid metal from relevant spatio-temporal patterns extracted from this data. The team will perform verification and validation of their work using physical modeling and technologies such as laser-Doppler vibrometers where possible. Modeling of the physical melt pool will also be used to guide and inform the experimentation.

Augmented Reality for Interactive Robot Control

Mentors: Alessandro Cattaneo and Troy Harden
Students: Levi Manring, John Pederson, Dillon Potts

Augmented reality (AR) is an emerging field that offers new tools for human-based control and interaction with complex systems. AR technology enables to overlay additional sensory information to the surrounding environment with the ultimate goal to alter human’s perception of the real world. As an example, off-the-shelf AR headsets offer the ability to map the surrounding environment and overlay holograms (i.e. three-dimensional images) on top of it. The overlaid holograms offer enriched visual information and can be used as a means for feedback control. Nowadays robots can fulfill complex tasks (e.g. manufacturing and assembling tasks) under demanding constraints (e.g. tight geometric tolerances, short processing times). Modern robots are effective in achieving these tasks when their jobs are planned in structured environments and unexpected contingency events are kept at minimum. For example, welding robots in an automated car production line are able to perform repetitive jobs with high accuracy on components that always occupy the same position relative to the robot. If obstacles are present (e.g. frame of the car), the welding robot arm trajectory can be planned ahead of time to minimize the risk of collisions (e.g. when reaching the inside of the car frame). This project explores the possibility of using AR for the interactive control of a Yaskawa® SIA5F 7‑axis articulated robot arm in non-structured environments. A state‑of‑the‑art Robotiq® 3-finger adaptive gripper mounted on the robot’s wrist offers broad flexibility in the shapes of the object that can be picked up.

Input Dimension Reduction for Materials Models

Mentors: Emily Casleton, Earl Lawrence, DJ Luscher, Saryu Fensin
Students: Samuel Myren, Andrew Shoats, Emilio Herrera

Computational models for representing the behavior of materials under stress are becoming increasingly computationally intensive and complex. Their increased computational burden means it is difficult to use them directly in problems of parameter estimation and uncertainty quantification, which require running the code many times. As a result, these models are often replaced by statistical approximations, called emulators, for these tasks. However, these approximations can be difficult to build when the materials model has a large number of inputs. This project will consider the use of active subspace methods to reduce the input dimensionality for a materials model and then calibrate the model with experimental data. We will focus on tantalum whose strength and heat resistance make it very useful in applications like armaments, jet engines, and spacecraft reentry. We will experiment and model tantalum single crystals placed under stress using an MTS machine to compress the crystals uniaxially.

Remote Detection of Abnormal Behavior in Mechanical Systems

Mentors: Eric Flynn and Adam Wachtor
Students: Greta Colford, Erica Jacobson, Kaden Plewe

In monitoring and surveillance applications, local measurements are often not possible due to instrumentation costs and/or access restrictions. In these situations, one must rely on remote measurements for making assessments about the system and detecting anomalous behavior. However, in remote monitoring situations, rather than capturing the system behavior directly, one can only capture the behavior after it’s been “filtered” through the propagation environment that stands between the system and the sensor(s). In addition, one must also necessarily capture the emissions of all other systems present in that environment, confounding the measurement.

The fundamental difficulty with remote detection then is that after the signal of interest has been attenuated by the propagation environment, it becomes undetectable among the other emissions. However, if the signal is continuous, or at least repetitive, then measurements that are in some way “averaged” over time can make the undetectable, detectable.

Accumulated Lifetimes in Single-Axis Vibration Testing

Mentors: Dustin Harvey, Colin Haynes, and Stuart Taylor
Students: Adam Bouma, Abigail Campbell, Thomas Roberts

Mechanical vibration environments are typically derived from the response of a system-level field test with a multi-axis excitation source. Most laboratory testing for system and component qualification is performed using multiple, single-axis tests. Ideally, components would be tested on the system in realistic, field boundary conditions, but this approach rarely occurs because of financial, hardware, and facility constraints. Often, these constraints also prevent individual components from being fixtured on the real system during testing, requiring the component to be tested in a different configuration than its intended design. As an added layer of complexity, components that were not instrumented in the field test will require an additional, intermediate system-level test to measure their specific component environments.

Characterizing Dynamics of Additively Manufactured Parts

Mentors: Garrison Stevens and Kyle Hammond
Students: Gary Adkins, Clayton Little, Peter Meyerhofer

Additive manufacturing (AM) is a rapidly expanding industry due to the capability to efficiently prototype and produce complex parts. However, we know that AM parts do not have the same mechanical properties as an identical part machined from bulk material. In fact, properties are likely to change based on the process parameters selected during manufacturing. For example, the build orientation and type of heat treatment used in selective laser melting have been shown to change the lattice structure of titanium parts, thus changing the part density and compressive strength (Wauthle et al. 2015). However, our understanding of the effects of manufacturing process on dynamic response remains lacking despite structural dynamics often being a critical system behavior of interest.

In this project, students will characterize the structural dynamics of geometrically identical parts additively manufactured using different process parameters. First, parts will be tested using a shake table while dynamics are measured using digital image correlation (DIC). DIC is an optical method that uses cross-correlation to track surface movements, providing full-field displacement and strain measurements. With advances in our data acquisition and digital camera technologies, we can collect high frame rate measurements to be used for DIC. Next, part dynamics will be simulated using a finite element model. Test analysis correlation of FE model predictions to DIC measurements will be used to determine necessary corrections to material model parameters to account for AM process parameters.

Lectures

Wavenumber Spectroscopy Techniques for Inspection using Laser Ultrasound

Eric B. Flynn
LANL, Space and Remote Sensing (ISR-2)

Scanning laser-ultrasound, defined as the use of an area-swept laser for exciting and/or sensing ultrasonic waves, is a fast-emerging technique for structural health monitoring (SHM) and nondestructive inspection (NDI). In this talk I will describe our implementation of three excitation-sensing modalities: laser-excitation, laser-sensing, and fully noncontact inspection, as well as two waveform types: broadband pulses and band-limited harmonics. Each of these implementations uses some combination of laser-Doppler vibrometry, a high power pulsed laser, and/or traditional piezoelectric transducers. For each modality, I will discuss the specialized three-dimensional (X, Y, & time) signal processing techniques we have developed as well as the advantages and disadvantages that we found both in measurement quality and in practical implementation. I will describe in detail our application of wavenumber spectroscopy, which is the analysis of spatial-spectrum information in order to characterize wave propagation mediums. Using these measurement modalities and signal processing techniques, we have effectively imaged a range of corrosion- and delamination-type defects. I will also present some preliminary results on the more challenging problems of detecting small defects, such as cracks, as well as fluid levels in containers. Finally, we conclude with a discussion on the role laser-ultrasound could play in a greater guided-wave hybrid SHM-NDI framework.

Structured Dynamics and Biology

Bridget Martinez
LANL, Engineering Institute

Mechanoreciprocity refers to a cells’ ability to maintain tensional homeostasis in response to various types of forces. Physical forces are continually being exerted upon cells of various tissue types, even those considered static tissues, such as the brain. Through mechanoreceptors, cells sense and subsequently respond to these stimuli. These forces and their respective cellular responses are prevalent in regulating everything from embryogenic tissue-specific differentiation, programmed cell death, and in what has garnered the most attention, disease progression. Abnormal mechanical remodeling of cells can provide clues as to the pathological status of many tissues. This becomes particularly important in cancer cells, where cellular stiffness has been recently accepted as a novel biomarker for cancer metastasis. Several studies have also elucidated the importance of cell stiffness in cancer metastasis, with data highlighting that a reversal of tumor stiffness has the capacity to revert the metastatic properties of cancer. Here we summarize our current understanding of extracellular matrix (ECM) homeostasis, which plays a prominent role in tissue mechanics. We also describe pathological disruption of the ECM and the subsequent implications towards cancer and cancer metastasis. In addition, we also highlight the most novel approaches toward understanding the mechanisms which generate pathogenic cell stiffness and also provide potential new strategies which have the capacity to advance our understanding of one of human-kinds’ most detrimental medical pathologies. These new strategies include video-based techniques for structural dynamics, which have shown great potential for identifying full-field, high-resolution modal properties, in this case as a novel application.

B61-12 Life Extension Program: System Qualification and Joint Testing

Curtt Ammerman
LANL, B61-12 Life Extension Program (Q-15)

Global Security Intelligence Work at LANL

Jon R. Schoonover
Deputy Associate Director
Threat Identification and Response

Los Alamos has done intelligence work since its inception. This talk discusses why LANL does Intel work and what that work is based on LANL’s approach to strategic deterrence and stockpile stewardship. The talk also discusses how structurally the global security part of LANL accomplishes Intel work. Opportunities for students and future staffing is also discusses.

Space Reactors

Patrick McClure
LANL, Systems Design and Analysis (NEN-5)

The Kilopower Project was initiated by NASA’s Space Technology Mission Directorate/Game Changing Development Program in fiscal year 2015 to demonstrate subsystem-level technology readiness of small space fission power in a relevant environment for space science and human exploration power needs. The Kilopower Project centerpiece is the Kilopower Reactor Using Stirling Technology (KRUSTY) test, which consists of the development and testing of a fission ground technology demonstrator of a 1 kWe-class fission power system. The technologies developed and validated by KRUSTY are extensible to space fission power systems from 1 to 10 kWe, which can enable modular surface fission power systems for human exploration, as well as higher power future potential deep space science missions. The KRUSTY demonstration is co-funded by NASA and the Department of Energy National Nuclear Security Administration (NNSA). The talk will focus on how the Kilopower project could provide power for potential missions on Mars or for deep space science missions. In addition, the talk will cover the recent KRUSTY series of tests performed at the Nevada National Security Site and their impact on moving the Kilopower project forward.

Making Effective Presentations and Writing a Conference Paper

Phillip Cornwell
Professor of Mechanical Engineering
Rose-Hulman Institute of Technology

The vast majority of engineers do not use an approach to presentation slide design that is supported by research on effective technical communications. The slides often contain bulleted lists that are designed to be useful for the presenter but are of virtually no value to the audience. In this talk, Dr. Cornwell will discuss how to give effective oral presentations using a type of slide design called the assertion-evidence approach. In this approach, the speaker explicitly identifies the messages (assertions) he or she is trying to present and provides the evidence to support these assertions. The information in this talk is based on the work of Dr. Michael Alley from Penn State. In addition to the main topic of effective slide design, Dr. Cornwell will also present an overview of conference paper requirements.

Aircraft Structural Dynamics

Nick Lieven
Professor of Aircraft Dynamics
University of Bristol

We explore the extreme dynamic behavior of aircraft and their related systems. A particular focus will be on the physical factors which lead to flutter and aircraft instability. Although flutter can be an entirely predictable phenomenon there is an increasing awareness that such aeroelastic instabilities can be caused by structural nonlinearities and human intervention. The talk will explain how these factors can interact in a potentially destructive way and what technologies and design modifications can be deployed to mitigate against this behaviour. The talk will consider both the modelling aspects of aircraft nonlinearities and the practical considerations associated with flight testing and – ultimately – flight safety.

Offshore Wind Farm Turbines Performance Monitoring and Wind Energy Current Trends

Nikolaos Dervilis
Dynamics Research Group, Department of Mechanical Engineering
University of Sheffield

The use of offshore wind farms has been growing in recent years. United Kingdom is presenting a geometrically-growing interest in exploring and investing in such offshore power plants as the country's water sites offer impressive wind conditions. However, the cost of an offshore wind farm is relatively high, and therefore their reliability is crucial if they ever need to be fully integrated into the energy arena.

The Structural Dynamics and Acoustic Systems Laboratory (SDASL)

Pete Avitabile
Professor Emeritus, Mechanical Engineering
University of Massachusetts Lowell

Co-Director, Structural Dynamics and Acoustic Systems Laboratory
Director, Modal Analysis and Controls Laboratory

The Structural Dynamics and Acoustic Systems Laboratory (SDASL) focuses on research related to analytical and experimental problems in the areas of structural and acoustic systems. The main thrust of the SDASL is to develop, employ and improve techniques to solve these problems using analytical approaches that are verified through experimental techniques.

Over 4 decades experience in design, analysis, finite element modeling and experimental modal and structural dynamic testing. Main area of research is structural dynamics specializing in the areas of modeling, testing and correlation of analytical and experimental models and integration of analytical and experimental techniques. Research, testing and consulting performed for automotive, aerospace, defense and computer/consumer related areas. Written over 200 technical papers and given numerous seminars in the areas of experimental modal analysis, structural dynamics, vibration fixture design, and modeling and correlation.

2017

Projects

Light-field imaging of structural dynamics by Advanced Computer Vision and Machine Learning

Mentors: Yongchao Yang and David Mascareñas
Students: Benjamin Chesebrough, Sudeep Dasari, Andre Green

Structures and systems usually have complex geometries, material properties, and boundary conditions, and exhibit spatially local, temporally transient, dynamics behaviors. Measuring their dynamic responses at multi-dimensional, high spatial and temporal resolution, are essential to accurately characterize the structure’s dynamics and validate the high-fidelity mathematical model (e.g., a finite element model). However, it is a significant challenge to obtain high-resolution vibration measurements using traditional techniques. On the other hand, digital video cameras are relatively low-cost, agile, and provide non-contact, high spatial resolution, simultaneous, measurements where every pixel effectively becomes a measurement point on the structure.

Recently, light field cameras are emerging as a new imaging tool, capable of capturing 3-dimensional, high-resolution information about the object in a single shot. Different from traditional digital cameras that measure only light intensity, light field imaging measures not only the light intensity, but also the direction of the light ray. Thus, they allow for reconstruction of the 3-dimensional, full-field object model, and perspective refocus on the object even after the images are taken.

The proposed research aims to develop a novel light field imaging method for capturing structural dynamics responses at 3-dimensional, high resolution, and full-field. The dynamics model will be physically connected to computer vision and machine learning models to effectively and efficiently extract such structural dynamics information from the light field measurements.

In this project, under mentors’ guidance, the students will develop a new structural imaging method for capturing 3-dimensional, full-field structural response. They will establish a 3D, high-resolution, full-field dynamic model from the light field measurements, using the latest computer vision and machine learning techniques. They will design and build different types of lab-scale structural models, and conduct dynamics testing experiments to validate the method. Specifically, they will apply their developed new light-field imaging method to study the 3-dimensional, full-field, dynamics properties of the structure under environmental variations at a resolution and detail that was not previously possible. At the end of the project, the students will have solid understanding and skills of the fundamentals of structural dynamic modeling and testing, system identification, signal/video processing, and machine learning. Furthermore, they will grow interests and knowledge about the state-of-the-art sensing and imaging technology from a practical engineering perspective. Finally, they will develop a suite of MATLAB codes of the developed algorithm based on open-source light field imaging measurement and algorithms, computer vision, and machine learning codes, and be trained to write scientific reports for publications in top-tier engineering journals.

Applying the Concepts of Complexity to Structural Health Monitoring

Mentors: Chuck Farrar and Alex Scheinker
Students: Travis Andrews, William Robert Locke, Brian West

This project will begin with the hypothesis that damage, regardless of the type, increases the complexity of system. The focus of this study will then be to define appropriate measures of complexity for SHM and to develop a framework whereby the effectiveness of a proposed damage-sensitive feature can be evaluated by its ability to quantify changes in complexity. The concept of complexity has been study extensively in many different contexts and a quick search on Google will reveal numerous measures that have been proposed for quantifying complexity. A portion of this study will be to assess these different measures and find the ones that are most applicable to detect a specific type of damage. It is anticipated that there will not be a single measure of complexity that is appropriate for all types of damage. The measure that might be appropriate to identify a crack in a vibrating plate may not be the same one that would be used when identifying corrosion in the same plate.

This project will investigate these measures of complexity analytically with numerical simulations and experimentally with various test structures subjected to simulated damage. An initial portion of the project will be to develop an analytical/experimental plan that can be accomplished under the time constraints of the program. Time permitting, the group can investigate the consistency of the complexity measures found to be best suited for a particular type of damage with the measures of complexity that are used by the continuum damage mechanics community.

Augmented Reality for Next Generation Structural Inspections

Mentors: Troy Harden, Alessandro Cattaneo
Students: JoAnn Ballor, Oscar McClain, Miranda Mellor

A primary concern for the Department of Energy mission is ensuring that critical infrastructure maintains its structural integrity over the course of many years. If the structural integrity of holding tanks, pipelines, pressure vessels or other key infrastructure is compromised the result could be the unintended release of dangerous waste or chemicals into the environment. The impact of such a release includes damage to the environment, health hazards for humans, and degradation in the public trust of the DOE. Unfortunately, to date the tools available to structural inspectors to perform structural inspections has been very limited. In many cases inspectors are limited to using their eyes, ears, a tape measure and a hammer to perform an inspection. Notes are taken using a pad of paper and a pencil. In some cases a tablet computer might be used to take notes along with a digital camera for documentation. Unfortunately these techniques are clumsy and result in very few data points being measured and recorded. In some cases thermal imagers or ultrasonic non-destructive testing techniques might be used, but these techniques are expensive, and bulky to deploy. In this work we will focus on developing novel structural inspection tools for environmental management applications based on emerging augmented reality technology. Augmented reality technology allows for holograms to be placed in the real world. It represents an exciting new way for inspectors to both collect, interact with, visualize, and analyze inspection data. For example, modern augmented reality devices typically come equipped with a depth imager. This imager can be used to make high-resolution 3D models of critical infrastructure on-the-fly during an inspection. This data could be used to precisely track how concrete beams or steel panels are corroding. A high-resolution 3D measurement of a corroded area could be made. If measurements are made year after year, inspectors will be able to instantly overlay the measurements from the prior year to visually compare how the structure has changed. A plot can be made on-the-fly showing how the volume of corrosion has changed. Furthermore, RGB images can be taken all over the structure and used to compare with prior year inspections. Eventually all this data can be fed to machine learning algorithms and finite element models to provide additional insight into structural integrity. Another major advantage of augmented reality devices is that they are typically hands-free, thus providing the inspector the use of their hands to perform tap tests or to navigate difficult terrain or to even better operate while rappelling off the side of a structure. Augmented reality itself could be interfaced with other sensors to not only record data, but to also overlay data from ultrasonic or thermal imagers onto the actual structure as desired in order to get a more complete picture of structural integrity.

Optimal Control for Single-Axis Vibration Testing

Mentors: Stuart Taylor and John Heit
Students: Vivian Cai, Michael Maestas, Alan Williams

In mechanical testing for system qualification, test engineers often face the challenge of representing a multi-axis vibration environment using a single-axis shaker table with only one degree of freedom for control. On large, complicated systems, the target environments are often defined as power spectral densities (PSDs) at multiple locations. These PSDs are almost invariably defined such that they are not physically realizable in any boundary condition. If one location responds with exactly its target PSD, another location will respond with a PSD that is different from its respective target.

Test engineers have developed various strategies for compromise control schemes that aim to reduce the extent to which the response PSD of any given location fails to match that of its target PSD. However, most practical approaches that have been implemented are not optimal, do not adjust for nonstationary behavior in the system, and make no attempt to reduce the overall error by improving the controllability of the system under test.

In this project, students will learn the relevant linear system, control, and signal processing theory underpinning mechanical testing for random vibration environments. Students will propose control strategies to minimize the error between specified target PSDs and measured response PSDs for a given test article. Students will program and implement their proposed strategies to work in concert with standard random vibration controllers. Students will implement their control strategies in a rapid prototyping and development cycle, troubleshooting and refining the strategies to select the best one based on objective benchmark testing. They will report their findings in a conference paper to be presented at the 36th IMAC Conference and Exposition on Structural Dynamics.

Reduced Order Quantities of Interest for Engineering Quantification of Margins and Uncertainties

Mentors: Garrison Stevens and Emily Casleton
Students: David Alexander, Polina Alexeenko, Bridget Daughton

Modeling and simulation plays a major role in the design and safety assessment of complex dynamic systems. Quantification of Margins and Uncertainties (QMU) seeks to evaluate uncertainties in model predictions to ensure a safe margin between an engineering system’s behavior and a prescribed failure regime. Within QMU, uncertainty quantification requires a comparison of predictions from a simulation model to experimental physical measurements of selected quantities of interest (QoI). As systems become more complex, a single QoI may span across several operational settings (i.e. temperatures or frequencies) as well as over large spatial and temporal domains. Increasingly, experimental techniques are available to collect full-field measurements across these domains. However, benefits of detailed experimental data cannot be fully realized until calibration and validation metrics are expanded to capture the same full-field, multi-domain behaviors.

The goal of this project is to identify reduced order quantities of interest that combine measurements and simplify predictions over large domains to a metric feasible for implementation in QMU. These metrics should reduce the dimensionality of a measured or simulated physical quantity while maintaining relevant correlations across the domain. Metrics should also be informative for model calibration, effectively reducing uncertainty in model parameters.

In this project, you will conduct a thorough analytical study to compare various statistical decomposition techniques for reducing the dimensionality of full-field, multi-domain QoI. You will collect full-field time histories of structural vibrations at varying excitation frequencies as well as develop a finite element model of the system for predicting the vibration response. You will apply the statistical decomposition to measured and simulated quantities and compare the model predictions and the true system to calibrate uncertain parameters. A comparison of the reduced order QoI will be completed and validated with known material properties of an aluminum plate. Finally, the best performing reduced order QoI will be extended for uncertainty quantification of a composite plate model.

Monitoring Dynamic Systems through Remote Sensing

Mentors: Eric B. Flynn, PhD & Dustin Y. Harvey, PhD
Students: Peter Fickenwirth, Charles Liang, Tyrell Rupp

In monitoring and surveillance applications, local measurements are often not possible due to instrumentation costs and/or access restrictions. In these situations, one must rely on remote measurements for making assessments about the system and detecting anomalous behavior. However, in remote monitoring situations, rather than capturing the system behavior directly, one can only capture the behavior after it’s been “filtered” through the environment that stands between the system and the sensor(s). In addition, one must also necessarily capture the emissions of all other systems present in that environment, confounding the measurement.

Typically, one develops remote detection algorithms by training on remote measurements of the activity of interest. However, in many cases, the activity of interest is so rare that training measurements in the remote environment do not exist, and one only has local laboratory measurements to train-on. On the other hand, one may have both local and remote measurements of other, more common activities.

This study aims to develop a new approach for building remote anomaly detection algorithms by utilizing both local and remote measurements of normal activity in order to learn how the propagation environment affects the measurement.

The test case for this study will be remote condition monitoring of rotating machinery. Downtime of critical machinery can be expensive and pose a risk to safety. For example, failure in the cooling machinery in a power plant can have devastating consequences. In facilities where dozens of machines may be operating simultaneously, thorough instrumentation of every machine may be logistically and economically infeasible. A system that could monitor multiple machines remotely from a central measurement instrumentation location would be incredibly valuable for aircraft monitoring, automobile monitoring, smart homes, smart power grids, assembly lines, process control, manufacturing, and power generation.

Other broad areas that will benefit directly from this research include structural health monitoring, disease detection, emissions treaty verification, nonproliferation, seismology, and cyber-physical security.

Structural Health Monitoring of Additively Manufactured Parts Using Fiber Bragg Gratings

Mentors: Scott Ouellette and Michelle Lockhart
PI: Alexandria Marchi
Students: Carl Fauver, Elon Gordon, David Petrushenko

Many industries have begun exploring the numerous opportunities offered by replacing traditional manufacturing techniques with additive manufacturing (AM). Aerospace, automotive and medical companies have embraced the advantages of AM including mass customization of functional, low-volume, highly complex, multi-material geometries. While traditionally manufactured parts involve forging and casting of bulk materials to develop the final part, additively manufactured parts are produced by continuously layering lines of the bulk material to build up the object, e.g., the fused deposition modeling process (FDM). The FDM process is subject to a high degree of variability, producing parts with several defects and abnormal mechanical properties. Advocates of AM promote its unique ability to fabricate one-off parts; however, with part-to-part variation arises the need to monitor the structural health of each unique part. Embedding optical fiber Bragg grating (FBG) sensors into AM structures allows for measurements of internal strain and temperatures over the operational lifetime of a part. FBGs are preferable over strain gauges and thermocouples, in part, due to their resiliency to electromagnetic interference; which is of concern when implementing AM parts in defense applications. Traditionally, strain gauges are adhered to the surfaces of the parts of interest because the maximum strain in bulk, non-laminar, structures is at the surface. Embedding FBGs is necessary because of the laminar behavior of AM parts coupled with their potential use in extreme environments where the surfaces of parts may be compromised over time. Keys to fully capturing the mechanical properties of a part include full adhesion of the FBG to the part and comprehensive understanding of the thermal and mechanical sensitivities of the FBG.

In this project, you will model given part geometries to determine proper locations to place FBGs. You will calibrate FBG sensitivities against thermal fluctuations and known strains. You will develop a technique to properly embed FBGs into AM parts and experimentally test the mechanical integrity of parts with and without FBGs, validating your models.

You will learn the AM process using a Lulzbot Taz to 3D print your geometries. Micron Optics instrumentation will be provided for use in interrogating the FBGs. You will use ABAQUS to model the various geometries and loading conditions, which will be compared to your experimental testing setup. LabVIEW software and/or Python code scripts will be used for data acquisition.

Lectures

Tom Paez - Random Vibration Analysis: What it is and what it's good for?
Phil Cornwell - Making Effective Presentations
Garrison Stevens - Multi-Physics Modeling
Jim Wren - Thinking Telescopes
Nikos Dervilis - Wind Turbines
David S. Moore - Explosives
Amy Regan - Satellites
Nick Lieven - Aerospace Structures
Curtt Ammerman - B61 Lifetime Extension Program
Yongchao Yang - Full Field Imaging of Structural Dynamics
Alex Marchi - PU Measurements
Reinhard Friedel - LANL Satellites

2016

Projects

Enhancement of an Acoustic Sensor System for a Spin-down Damage Assessment Air Bearing

Mentors: John Heit, Alessandro Cattaneo
Team: Nicholas Diskerud, Wesley Scott, Martin Ward

Non-destructive evaluation (NDE) techniques for inspection and damage assessment of expensive and complex mechanical systems are highly sought after in the Engineering community. One such NDE technique that was developed for the aerospace industry involves spinning objects on an isolated air bearing device and observing the rate of angular deceleration, also known as a “spin-down” test. Air friction on the outside of the object provides a breaking mechanism to dissipate rotational kinetic energy and cause the object to decelerate. If components within the object were loosened or broken, additional energy would be dissipated during rotation in the form of frictional heat and the object would decelerate at an increased rate. The air bearing spin-down method is extremely sensitive and can detect small amounts of mechanical damage such as a broken wire or loosened powder.

While the spin down rate of an object can be used to assess the presence of mechanical damage, it can not necessarily specify the location of the damage. To overcome this limitation, a 2015 Los Alamos Dynamic Summer School (LADSS) engineering team developed a standalone acoustic sensor system which attempted to locate mechanical impacts from broken parts within a rotating system. This system was extremely limited in its capabilities as it suffered from low signal-to-noise levels and limited signal processing.

To overcome these limitations, a higher fidelity acoustic system that employs acoustic sensors with integrated pre-amplifiers, was procured. In this project, you will utilize this acoustic sensor hardware to develop a measurement system and methodology to detect, locate and identify mechanical damage inside a spinning test unit. This will include developing smart procedures for the placement and mounting of the acoustic sensors, implementing signal processing algorithms to extract information from the time and/or frequency domain data, and ultimately assess the capabilities and limitations of the acoustic sensor system.

Experimental Investigation of Numerical Calibration Techniques

Mentors: Kendra Van Buren and François Hemez
Team: Philip Graybill, Eyob Tarekegn, Ian Tomkinson

In engineering disciplines, it has become accepted practice to develop numerical models to investigate the behavior of systems. For example, standards have been developed by professional societies to regulate practices for developing reliable numerical models of buildings, cars, and mechanical structures. Numerical simulations offer a cheap alternative to the design-build-test paradigm to explore new design concepts. One issue, however, is that parameters of a numerical simulation are often unknown or uncertain, and as such, predictions of the model are uncertain. Small-scale experiments can be performed to mitigate this issue, for example carrying out tensile tests to determine the Young’s modulus of a material, which can be used to inform a larger-scale numerical model. Numerical models are then used to forecast (predict) the behavior of the system at different settings. Forecasting can, at times, exercise models “far away” from conditions where they have been developed and calibrated. This calls into question the quality (and veracity) of predictions hence obtained. While the practice of calibration is omnipresent in simulation sciences, the danger of extrapolating far from calibration conditions is not always appreciated, and there is a lack of rigorous methodology to shed light on this potential challenge.

The objectives of this project are to, first, develop an example that illustrates the dangers of calibrating in one domain and forecasting in another domain and, second, start formulating a general-purpose framework to study the extent to which this could be an issue. The application is to predict the amount of payload that a spring attached to a quad-engine copter can withstand before failure (it is the forecasting domain). Small-scale experiments performed to exercise the spring will be used to calibrate the model. The project will develop simple phenomenological models, perform physical experiments, calibrate the models, integrate them into a forecasting simulation, and assess the accuracy of forecasts obtained in conditions that differ from those encountered during calibration. Technical skills acquired by working on this project will include performing vibration testing (modal analysis), developing simple finite element models whose predictions will be validated using vibration measurements, and becoming familiar with calibration methods.

Feedback Control of Modulated Inertial Generators for Energy Harvesting Applications

Mentors: Scott Ouellette and Antranik Sirinosian
Team: Rachel Gaspar, Matthew Mascareñas, Chriss Sanpakit

In recent years, increased interest in broadband vibration energy harvesting (VEH) schemes has been a main topic of interest among researchers. One of the most successful approaches towards broadband vibration energy capture has been with bistable inertial generators. These devices leverage a nonlinear restoring force to exploit the hardening spring response to increase the resonant frequency bandwidth beyond the characteristically narrowband resonant frequency associated with conventional linear inertial generators. However, one issue with bistable energy harvesters is the presence of low-amplitude oscillations whose energy is insufficient to overcome the potential energy separatrix barrier between the competing potential wells. One method of overcoming the low-energy orbits is to modulate the separatrix barrier height by means of an external controller. In this design, a permanent magnet affixed to the tip of the inertial generator interacts with an electromagnet to generate the nonlinear restoring force. Recent studies have successfully demonstrated the augmented broadband resonant response of this type of design; however, at present the energy cost for operating the electromagnet ultimately negates the benefits of the modulation scheme.

In this project, you will develop an intelligent feedback control scheme for the electromagnet with the goal of enabling a net-positive energy capture. You will apply a meta-model of the nonlinear oscillator system subject to this feedback-control scheme to verify the augmented broadband response, as well as experimentally test the control scheme.

Example feedback control schemes of interest in this investigation are: gain-scheduling, extremum-seeking, and linear optimal control. Time-permitting, a comparative investigation among multiple control schemes may be conducted for a more exhaustive study. A mini-project consisting of determining the system damping as a function of magnet spacing (due to Faraday effects) will be used to acclimate you to the data acquisition hardware and software.

You will be using Python for the numerical simulations and experimental data post-processing. For hardware control and data-acquisition, you will be using LabVIEW. The experimental apparatus will consist of an electro-dynamic shaker along with a laser Doppler vibrometer for measuring tip displacement of the inertial generator.

Identification of High-resolution, Full-field Dynamic Loads on Structures by Computer Vision and Unsupervised Machine Learning

Mentors: David Mascareñas and Yongchao Yang

Team: Alex Roeder, Lorenzo Sanchez, Huiying Zhang

Real-world structures such as civil structures and aerospace structures are subjected to various dynamic loads, e.g., traffic, earthquakes, hurricanes, and aerodynamic loads, which are spatially local and distributed on structures. Assessment of operational performance, prediction of the dynamic responses, and prognosis of the remaining service life of the structure therefore require accurate, high-resolution measurements and modeling of these dynamic loads acting on the structure, which is extremely difficult, if not impossible, in the current state of the art and practice. The proposed research aims to develop a novel method for identification of the high-resolution, full-field loads to the structure, by using the digital video camera measurements and exploiting the recently-developed advanced computer vision and unsupervised machine learning techniques. The non-contact, remote, simultaneous sensing capability of the proposed technique should enable truly high-resolution, full-field force estimation that was not feasible before.

In this project, the students will develop a new method for identification of the full-field dynamic load on the structure. They will establish a high-resolution, full-field dynamic model using high-resolution, high-speed video camera measurements and the latest computer vision and machine learning technology. They will design and build a lab-scale bridge model, apply traffic and wind loads on the structure, and conduct experiments to validate the method. At the end of the project, the students will have solid understanding and skills of the fundamentals of structural dynamic modelling and testing, system identification, signal processing, and machine learning. They will develop a suite of MATLAB codes of the developed algorithm based on open-source computer vision and machine learning codes.

Multi-Source Sensing and Analysis for Machine-Array Condition Monitoring

Mentors: Dustin Harvey and Eric Flynn
Team: Shannon Danforth, Jaden Martz, Alison Root

Downtime of critical machinery can be expensive and pose a risk to safety. For example, a failure in the pump machinery that runs the liquid cooling systems for LANL’s high-performance-computing facility could lead to computing downtime and damage to expensive computing hardware. In facilities such as LANL’s HPC center, where dozens of machines may be operating simultaneously, thorough instrumentation of every machine is logistically and economically infeasible. In many cases, such a large, complex instrumentation network might require its own monitoring network. A system that could monitor an array of machines remotely from a central measurement instrumentation location would be incredibly valuable to numerous industries.

Most machine condition monitoring research focuses solely on one type of measurement, such as vibration. However, in many applications, other measurement streams are either already available or can easily be made available. These other streams include voltage and current on the common power line, airborne acoustics, and electro-magnetic fields. How to combine these multi-source data streams into a single analysis is an open-ended problem.

The focus of this research will be in answering three important questions:

How can heterogeneous measurement data streams (i.e. different types, different sample rates, etc.) be combined and analyzed together as a whole?
Among an array of machines, can individual machines be identified and monitored without individual instrumentation by utilizing multi-source measurements (vibration, current, voltage, acoustics, and EM) from a central instrumentation point?
In multi-source collection and analysis, can one assess the value of individual data streams?

These questions will be answered by constructing a bench-top rotating machinery array, developing a centralized multi-source instrumentation package, performing controlled tests involving abnormal machinery behavior, and developing & applying new signal analysis tools. If successful, the team will have the opportunity to make measurements of in-service machinery at the Laboratory.

The ability to perform large-scale monitoring using a single, information-rich measurement point would be significant for a number of applications including aircraft monitoring, automobile monitoring, smart homes, smart power grids, assembly lines, process control, manufacturing, and power generation.

Self-Sensing 3D-printed Parts with Designer Mechanical Characteristics

Mentors: Alexandria Marchi and Adam Wachtor
Team: Ben Katko, Derya Tansel, Jennifer Yasui

Aided by the lowering costs of open-source fused deposition modeling (FDM) machines, once thought of as a simple toy for hobbyists, Additive Manufacturing (AM), also known as 3D-printing, is now quickly becoming a formidable and promising manufacturing process for the engineering community. Complex geometries and comprised of multi-materials can be shaped constructed with AM (Fig. 1) in ways that are too costly, time consuming, or difficult impossible to fabricate using traditional manufacturing methods. For a Design Engineer to take full advantage of AM, they must be able to design and qualify parts with desired properties. This project will focus on developing a sensing system for monitoring the quality and predicting the mechanical properties of additively manufactured parts using self-sensing structural materials.

Time Dilation in Qualification Testing

Mentors: Stuart Taylor & Colin Haynes
Team: Milo Prisbrey, Jacob Senecal, Manik Sethi

Often in mechanical testing for system qualification, there is a mismatch between the service environment and the testing environment in terms of the test duration or severity. When a mismatch exists, it is usually because the time available for testing is much shorter than the service lifetime, or the facility available for testing has insufficient capability for reproducing the mechanical insults expected during the service life. Test engineers have adopted common approaches for circumventing these shortcomings. In the former case, a common approach is to increase the severity while shortening the duration, using fatigue analysis. In the latter case, a common approach is to consider the response of a surrogate structure, using the shock response spectrum (SRS), to determine an equivalent testing environment that is within the capabilities of the testing apparatus. In all approaches, a damage model is implicitly or explicitly assumed.

In this project we will explore the application of material-specific damage models for the case of a deficiency in the capability of the test facility to apply the level of severity expected in the service environment. The events of interest are usually very short in duration and have high mechanical input levels compared to the bulk of the expected service life. Example events include rocket stage separation shock, airplane landing shock, and rail car linking. Using a material-specific damage model, we will develop an equivalence between an expected, but unachievable service environment and discrete repetitions of an achievable test environment. We will validate our approach using experiments with simple materials and compare the results to an industry-standard, SRS-based, material-agnostic approach.

Lectures

Scott Ouellette - Energy Harvesting
Stuart Taylor - Wireless Sensor Nodes
Eric Flynn - Ultrasonic Non-Destructive Evaluation
Jim Wren - Thinking Telescopes
Chris Stull - Information Gap Analysis
Yongchao Yang - Video Motion Magnification
Kendra Van Buren - Multi-intelligence Data Analysis
Nick Lieven - Aerospace Systems
Matthew Smith - Inland Waterways
Amy Larson - Computer Vision
Garrison Stevens - Verification and Validation
Reinard Friedel - LANL Satellites

2015

Projects

Developing Conservative Mechanical Shock Specifications

Mentors: Stuart Taylor and Dustin Harvey
Team: Matt Baker, Kelsey Neal, Katrina Sweetland

Many conventional approaches to specifying mechanical shock environments rely on the shock response spectrum as a measure of the severity of the environment. In this project, students will explore a variety of measures for the severity of a shock environment and identify measures, or transforms of those measures, that can uniquely specify a test environment to reproduce a desired severity level. These measures shall be restricted to those that can be observed in real time during a laboratory test, and they should be restricted to those that can be controlled in real time during a laboratory test. Students will conduct an analytical study based on measured shock data from a structure’s service environment and implement the chosen measures experimentally as a proof of concept.

Enhanced Spin-Down Diagnostics for Nondestructive Evaluation of High-Value Systems

Mentors: Colin Haynes and John Heit
Team: Joshua Pribe, David Sehloff, Clark Shurtleff

Detecting small changes in high-value systems that may indicate the onset of damage is a common and challenging problem. One nondestructive evaluation (NDE) technique for such systems involves spinning them on an isolated bearing device and observing the rate of angular deceleration – a “spin-down” test. If an increase in the angular deceleration is observed compared to that of systems in a healthy state, the additional energy dissipation is interpreted as evidence that damage has occurred.

While they can be very sensitive indicators of damage, spin-down tests are inherently limited in the information they can provide about any potential damage beyond its mere existence. We are interested in investigating additional diagnostics for enhancing the damage characterization capabilities for systems undergoing spin-down testing. In particular, information on the location and type of the damage in the test article are of interest.

The selection of sensing technology for these enhanced diagnostics is still an open question. You will have the opportunity to think about the physical phenomena and look into the methods that seem most promising to you. Both non-contact sensing and embeddable systems are options for a spinning system, and each has its own advantages and disadvantages. Laser velocimetry of the skin, magnetic flux sensors, and microphones are all examples of noncontact sensing systems that may be able to detect damage occurring inside a specimen. For embeddable sensing, using a data logger to record inputs from either accelerometers or ultrasonic transducers is likely to provide good sensitivity and localization capability.

You are also encouraged to consider techniques that will take advantage of the fact that the test article is spinning. When the article is spun about a horizontal axis, the orientation of gravity may cause the defects to be evident at particular angles of rotation. Monitoring the periodic and/or aperiodic, damage-related events as the article spins may yield a wealth of information not otherwise available through spin testing.

In this project, you will develop a measurement system and acquire data on a surrogate spin test unit. You will then implement signal processing algorithms to extract information from the data on the presence, location, and character of simulated defects.

Extending Human Proprioception to Cyber-Physical Systems

Mentors: Alessandro Cattaneo and Heidi Hahn
Team: Leah Dickstein, Kevin Keller, Ethan Robinson

In the last decade great advances have been made in the field of cyber-physical systems. Aerial robots have gone from tools primarily used by governments, to toys anyone can purchase on Amazon. Ground-robots have proven their worth performing dangerous tasks such as disposing of unexploded ordinance. Robot-assisted surgery has entered medical practice. Smart phones have become ubiquitous and have made experiences such as “getting lost,” almost an anachronism. The advent of low-cost rapid prototyping (e.g. 3D printers), extremely cheap, powerful, and most importantly easy-to-use, embedded computers, and the wide availability of low-cost, sensors and electromechanical components is only going to increase the role of cyber-physical systems in our daily lives. Despite these advances will lack intuitive interfaces for controlling and commanding cyber-physical systems. One of the best cases of this is the interface used to control and monitor commercial ground robots. These robots currently are controlled using something resembling a common video game controller. Furthermore, in order to know the state of the robot (e.g. arm joint locations, pose) the operator either needs to be looking directly at the robot, or they have to look at a screen that shows the state of the robot and it’s appendages. This situation is far from desirable. If we focus on the question of how we know the state of the robot we have to ask the question why this interface cannot be more intuitive. Humans have a proprioceptive sense that provides us information on how our bodies are distributed in space without having to look directly at our appendages. In this work we will explore using the phenomena of sensory substitution to build non-invasive, vibro-haptic interfaces that will allow us to extend the human sense of proprioception to cyber-physical systems. The ultimate goal of this work is to enable high-performance control of cyber-physical systems.

Extracting Spatio-Temporal Patterns from Motion Magnified video for Standoff Experimental Modal Analysis and Mass of Crop Root Systems

Mentors: Garrett Kenyon, Yongchao Yang, David Mascarenas
Team: Charlie Dorn, Tyler Mancini, Zack Talken

Experimental modal analysis has been a common experimental dynamics tool for structural engineers for more than three decades. This technique allows us to measure the natural frequencies and mode shapes of a structure. This information can be used to predict structural response to loads, and has also been applied to fields such as structural health monitoring. Typically performing an experimental modal analysis involves instrumenting a structure with accelerometers, exciting the structure with a modal hammer or electro-magnetic shakers, collecting the resulting data with a data acquisition system, and then going through a curve-fitting routine. This technique has had great success, but it can be time consuming and costly to complete a test. Recently techniques have been developed to magnify small movements in video. Initial experiments at MIT and Los Alamos National Laboratory suggest these video magnification techniques may be suitable for replacing the accelerometers used for experimental modal analysis with video cameras. The goal of this work is to see if spatio-temporal patterns indicative of mode shapes can be quantitatively extracted from video. If this work is successful it will help enable remote, non-contact, visual modal testing that is much less expensive and could possibly be applied all the way from the city-scale down to the scale of a single cell in the human body.

In-Process Ultrasonic Inspection of Additive Manufactured Parts

Mentors: Eric Flynn and Adam Wachtor
Team: Ian Cummings, Elizabeth Hillstrom, Rielly Newton

Additive manufacturing is taking over. Nevertheless, the greatest question still remains: “can I trust this 3D printed part?” Additive manufacturing (AM) introduces a new set of variables that affect manufacturing quality that simply are not part of traditional manufacturing processes. For example, in most traditional manufacturing processes, it is easy to qualify the solid base material from which a part is cut. In additive manufacturing, the base material is being built up, so for each part, one must qualify both the geometry of the part, as well its material construction.

The peculiarities and additional requirements for quality control of AM parts push the limits of existing non-destructive inspection techniques. AM usually involves small production quantities and each print run of the part is unique, so statistical sampling and destructive testing are not practical. Existing ultrasonic testing techniques, which are effective on relatively simple parts cut from a single block of material, are confounded by the internal intricacies that are possible with AM.

In this project, you will develop a new in situ ultrasonic inspection technique for additive manufacturing. You will design a system to actively probe the part with ultrasound throughout the entire build process. Resonant ultrasound spectroscopy (RUS) is one technique worth exploring. Acoustic resonance spectroscopy (ARS) and resonance inspection (RI) are similar techniques.

Resonance Control of RF Cavities

Mentors: Alex Scheinker and Alex Marchi
Team: Brian Page, Patricia Tan, Murray Orlando

Accelerating particles to billions of electron volts requires enormous electric fields, gradients of tens of millions of volts/meter. The only way to create such large fields without also immediately creating a destructive breakdown/discharge is to use fields that oscillate from millions to billions of times per second, thus never allowing the electrons in the conductor to accumulate long enough to cause a discharge. The most efficient way to create extremely high frequency oscillating electric fields is to use resonant cavities. Just like any mechanical system has many natural resonant frequencies of vibration, these structures can also support electromagnetic resonant fields, with frequencies inversely proportional to the inner radius of the structure, for a cylindrical cavity roughly f ~ 2.405c/r, where c is the speed of light and r is the inner radius. The resonant cavities used in particle accelerators operate in the radio frequency (RF) range: hundreds of megahertz to several gigahertz, and are extremely sensitive to mechanical deformation, which drastically changes the properties of the electric field, by directly influencing the resonant frequency. To successfully accelerate and bunch particles according to designed energies, the RF fields of accelerating structures must have precise amplitudes and must be precisely timed relative to a designed reference signal.

With careful machining and tuning a cavity can be created with dimensions such that it supports almost exactly a desired resonant frequency. Precise field control can also be utilized to time the signal so that it matches a desired reference signal’s phase and to maintain desired field amplitude. However, especially in superconducting structures, an interesting problem coupling the mechanical and electromagnetic properties of the resonant cavities causes the cavities themselves to distort and move their own resonance. As the field of a resonant cavity is brought on and reaches steady state (hundreds of microseconds), on the time scale of the mechanical system it looks like a step change in which an enormous field has suddenly been switched on, which kicks, pulls and pushes the cavity walls with an electromagnetic phenomenon known as Lorentz force detuning. This sudden kick of the cavity causes it to physically resonate at its mechanically modes, distorting the cavity geometry and moving the electromagnetic resonance. If this is uncompensated, from the RF point of view, even a tiny mechanical change (micrometers), corresponding to a ~200Hz, or 0.000015% resonance change in the case of a 1.3GHz cavity, is an unacceptable drift from the design value, causing the input RF power to reflect, the magnitude of the electromagnetic wave to drop and for a sudden and large phase drift of the cavity’s field relative to a desired reference signal, a change so large and fast that the RF feedback system cannot compensate to maintain the cavity field within design constraints. Typically, to deal with this problem, piezo electric elements are used in order to compensate for and cancel out the Lorentz force caused by mechanical cavity oscillations by an iterative learning approach.

Robust-Adaptive Monitoring of Structural Condition

Mentors: Kendra Van Buren and François Hemez
Team: Harry Edwards, Kyle Neal, Jack Reilly

In the discipline of Structural Health Monitoring (SHM), the condition of a structure is often assessed using a combination of measurements and numerical models that attempt to infer the structural state from changes in the vibration signature. A time series model, for example from the family of Auto Regressive and Moving Average (ARMA) models [1], is trained using the measured vibration responses of the pristine structural condition. The hypothesis is that the occurrence of damage manifests itself as a significant difference between what is measured on the (now damaged) structure and what the trained model predicts. This paradigm is one in pattern recognition for which methods have successfully been applied to real structures [2].

One challenge is the management of vibration response uncertainty. Vibrations of the structure change because external forces (wind loads for an aircraft, ground vibrations for a building, traffic on a bridge, etc.) vary, not just because the structure becomes damaged. Statistical methods, such as process control charts, have been developed to address the difficulty of environmental variability. Another difficulty is the modeling uncertainty. Several competing models of the vibration response can be available, which all fit the past observations equally well. Likewise, the parameters of a given model might need to tolerate some uncertainty in order to better capture the aforementioned environmental variability. Setting the values of model parameters too strictly leads to the risk that a difference between what the time series model predicts and what is measured on the structure is (erroneously) attributed to structural damage.

This project proposes to treat the challenges posed by the variability in the environment, unknown model forms and/or unknown model parameters as a problem in decision-making for which gaps in our knowledge, which are the previously listed sources of uncertainty, need to be managed robustly and adaptively to avoid false-positives or false-negatives of the structural condition. The proposal is to integrate three relatively well-known technologies and assess whether this novel algorithm performs better than more “conventional” approaches. Validation would involve performing vibration tests on a laboratory structure to collect data representative of the pristine and damaged states of the structure, and assess how the algorithm performs.

The first technology needed is a time series algorithm to model the vibration response. There is no need to develop our own technique; we can, instead, simply adopt a method suggested by experts in the field [2]. This algorithm will forecast the vibration signature, compare predictions to actual measurements, and synthesize these responses in a to-be-defined damage indicator.

The second technology is a description of uncertainty present in the problem, which will likely come as environmental variability, unknown model forms, and/or unknown model parameters. For simplicity, the description of uncertainty can be restricted to intervals defined around the nominal values. Probability laws, if available, can also be used. An analysis of robustness is then performed to propagate this uncertainty through the time series algorithm and obtain the resulting bounds of uncertainty of the damage indicator. Both uncertainty description and robustness will be implemented within the framework of info-gap analysis [3]. Info-gap software, which essentially consists of solving numerical optimization problems, will be made available in MATLAB® for integration with the time series algorithm.

In the info-gap theory of decision-making, the “size” of the uncertainty space is defined by a parameter labeled α. A larger value of α means, by definition, that more uncertainty can be tolerated while guaranteeing that the damage indicator remains below a critical level that would signify the occurrence of damage. The larger the value of α, therefore, the better because the decision reached in the analysis, such as “no damage”, is more immune to uncertainty present in the environment and model form. The third component, which is more research-oriented than the other two, is to actively learn what an appropriate value of α ought to be.

Using the value of α derived from previous times to analyze the vibration response at a future time is not necessarily a good strategy. A value of α that has worked in the past might not be appropriate to represent the size of the uncertainty space if the time series model no longer characterizes the current state of the structure. This might lead to unwanted “surprises.” We suggest to track and adapt α in time using a technique recently proposed to learn the values of model parameters [4]. The algorithm will need to be programmed to learn, not the parameters of the time series model, but the value of α as a function of time. Our hypothesis is that examining how α changes in time might lead to superior diagnostics of structural damage as compared to only monitoring the damage indicator.

The project will require some programming, not so much to develop the individual algorithms, which we are hoping to obtain from their original authors, but to interface them together. It will also require vibration testing of a to-be-defined laboratory structure to collect time-history data that characterize several conditions, from pristine to damaged structural states. Because the condition monitoring method proposed should be immune to environmental variability as much as possible, vibration tests will be executed at different levels of background “noise” and/or applied excitation. Results will be documented in a technical manuscript to be presented at the 34th International Modal Analysis Conference in February 2016.

Presentations

Automated Feature Design for Time Series Classification by Genetic Programming

Speaker: Dustin Harvey, LANL

Increasingly, researchers, engineers, and analysts find themselves inundated with numeric sequence data for use in classification tasks. Ever-decreasing costs of sensing, data acquisition, and data storage hardware have led to many applications where sequential data is abundantly available, but domain knowledge is very limited. Domain knowledge primarily drives the selection and design of pre-processing steps, features, models, and learning algorithms that constitute a pattern recognition system. These decisions are critical to the ultimate performance of the system, but few general guidelines or frameworks exist to guide the inexperienced practitioner. In this work, a genetic programming based approach called Autofead is proposed and demonstrated to automate feature design for numeric sequence pattern recognition systems. Autofead represents the first such system to leverage the power and efficiency of both numerical optimization and standard pattern recognition algorithms. Potential applications include the monitoring of electrocardiogram signals for indications of heart failure, network traffic analysis for intrusion detection systems, vibration measurement for bearing condition determination in rotating machinery, and credit card activity for fraud detection.

Deep, Sparse Representations of Form, Depth and Motion

Speaker: Garrett Kenyon, LANL

Sparse predictive coding modules have emerged as viable candidates for defining a universal cortical processor. Sparse autoencoders are self-organizing and can explain many of the linear and nonlinear response properties of simple cells in the primary visual cortex. Moreover, sparse predictive coding modules can be strung together into essentially arbitrary topologies. Here, we demonstrate how sparse predictive coding modules can be used to represent form, motion and depth features in a manner that enables subsequent analysis. We use an open source, high-performance neural simulation toolbox called PetaVision. In a typical simulation, we use either single images, sequences of video frames, or stereo image pairs, which are loaded into the input layer. A layer of cortical neurons then learns an optimal set of features for representing that input as accurately as possible while using as few active elements as possible, a process than can be simulated using lateral synaptic inhibition. The same process is then repeated to learn the receptive field properties of subsequent layers arranged in a hierarchical sequence. We then test the resulting sparse representations 3 ways. First, we test for the ability of a multi-layer sparse hierarchy to support good classification performance on object detection tasks, thereby assessing how the sparse representation represents form in a partially viewpoint invariant manner. Second, we test for the ability of a multi-layer hierarchy trained on short video sequences to enable good discrimination between different types of human actions, thereby assessing the how our sparse representation of motion enable better discrimination of spatiotemporal patterns. Third, we test for the ability of a sparse representations trained on stereo image pairs to reconstruct accurate depth maps. Our results illustrate how sparse predictive coding can be applied to a range of visual processing modalities and thus support the hypothesis that sparse coding can be used to define a universal cortical processing module that can be configured into arbitrary topologies for solving different information processing tasks.

Design Thinking

Speaker: Heidi Hahn, LANL

Design Thinking is a relatively recent approach to engineering for product design. Originally conceived by the commercial company IDEO, it has since become a major part of the curriculum at the Stanford University design school, which has further developed tools and techniques for implementation. This talk will provide an overview of the Design Thinking process. An example implementation – a project executed by Systems Engineering Masters students at the Naval Postgraduate School for LANL – will be used to illustrate how Design Thinking works in practice. Lessons learned about the practical application of Design Thinking will also be shared.

Dynamic Response/Full Field Stress-Strain from Limited Measurements

Speaker: Pete Avitabile, UCSD

Dynamic response due to operating and accessional loads is an important consideration in the design of structural systems. Fatigue and life usage are of great importance for structural health monitoring. However, in the design of many structural systems, the actual loading and structural condition (boundary condition, environmental condition, etc.) are not readily known or easily determined. These are critical pieces of information necessary for the design of any structural system.

Much effort has been expended by many researchers in attempting to identify these loading scenarios. At best, the forces and actual boundary conditions are approximate and have an effect on the overall predicted response and resulting stress-strain that is identified for subsequent evaluation. In addition, the operating system is generally only measured with a very sparse array of transducers to identify the actual loading conditions.

Recently several new approaches have been developed that allow for limited sets of measured data, in conjunction with a finite element model, to be used for prediction of full-field response. The limited sets of measurements are used with a unique expansion algorithm to obtain this full field information. The finite element model mass and stiffness matrices are used to obtain the normal constitutive relations as well as the modal characteristics. This information is used to develop the expansion algorithm and for the stress recovery during the back substitution process typically employed. A basic discussion of the process used is presented along with some results from a wind turbine system.

Algorithms for Active Learning

Speaker: Sanjoy Dasgupta, UCSD

The "active learning" model is motivated by scenarios in which it is easy to amass vast quantities of unlabeled data (images and videos off the web, speech signals from microphone recordings, and so on) but costly to obtain their labels. Like supervised learning, the goal is ultimately to learn a classifier. But the labels of training points are hidden, and each of them can be revealed only at a cost. The idea is to query just a few labels that are especially informative about the decision boundary, and thereby to obtain an accurate classifier at significantly lower cost than regular supervised learning.

There are two distinct ways of conceptualizing active learning, which lead to rather different querying strategies. The first treats active learning as an efficient search through a hypothesis space of candidates, while the second has to do with exploiting cluster or neighborhood structure in data. In this talk, I will show how each view leads to active learning algorithms that can be made efficient and practical, and have provable label complexity bounds that are in some cases exponentially lower than for supervised learning.

I will also give a brief overview of the other primary research interests of my group: hierarchical clustering, fast nearest neighbor search, and adaptation to the intrinsic dimension of data.

2014

Team Projects

Program Agenda

Characterization and Prognosis of Multirotor Failures
- Mentor: Dustin Harvey
- Students: Jordan Thayer, Jesse Coffey, Joe Brown
Development of Novel Human-Computer Interfaces for an Interactive Electronic Work Control System for Glovebox Operations
- Mentors: David Mascarenas
- Students: Hannah Ross, Kyle Embry, Andrea Hengartner
Drill Vibration Reduction through Active Absorption Part 2
- Mentors: Eric Schmierer, Alex Scheinker
- Students: Nick Martinez, Michelle Gegel, Jermaine Chambers
Graphene Tamper Resistant Seals
- Mentor: Alessandro Cattaneo
- Students: Jason Bossert, Christian Guzman, Axel Haaker
Laser Ultrasonics for Hybrid Structural Health Monitoring
- Mentors: Eric Flynn, Will Warren
- Students: Kyle Brown, Adam Gannon, Elizabeth Wheeler
Mechanical Shock Environment Synthesis and Experimental Testing
- Mentors: Stuart Taylor, Kendra Van Buren
- Students: Kaitlyn Kliewer, Greg Naranjo, Cassidy Fisher
Multimodal Sensing Strategies for Detecting Transparent Barriers Indoors from a Mobile Platform
- Mentor: David Mascarenas
- Students: Kaleb Kleine, Isaiah Acevedo, Dustan Kraus

Speakers

Doug Adams
Chris Anderson
Steven Anton
Pete Avitabile
Scott Backhaus
Alessandro Cattaneo
Chuck Farrar
Ed Fenimore
Eric Flynn
Steve Girrens
Dustin Harvey
Francois Hemez
Bill Hodgkiss
Nick Lieven
Dan Lofaro
DJ Luscher
David Mascarenas
Bill Priedhorsky
Alex Scheinker
Eric Schmierer
Stuart Taylor
Jim Wren

Projects and Speakers from previous year's programs

2024

Design of a Tuned Vibration Absorber with Friction Contact for High Location Ratio Application

Crack detection and localization optimization in Bi2Te3 wafers

Evaluating the Effect of Shaker Placement Optimization Priorities on Multi-Axis Test Results

Capture of Thin Film Vibromorphology using Laser Vibrometry and Frequency-Domain Viscoelastic Vibroacoustics

Selection of Optimal Reduced-Order Model of a Device Under Test Using Bayesian Classification

Extrapolating Dynamic Transfer Functions from Multi-Input Multi- Output Vibration Testing and Simulation

Characterization of Rayleigh-Taylor Instabilities Induced by Volumetric Energy Deposition

2023

A Characterization of the Uncertainty in Force-Control Testing for Aerospace Applications

Computerized Acoustic Resonance Crack Identification in Thermoelectric Bi2Te3 Wafers

Evaluating Vibration Controller Performance in Virtual and Hardware Tests

Adaptive Radio Frequency Target Localization

Quantitative Comparison of Vibration Testing Methods

Identification of Bird Species in Large Multi-channel Data Streams Using Distributed Acoustic Sensing

Generating and Quantifying Initial Conditions for Volumetric Energy Deposition Driven Rayleigh-Taylor Instability Experiments

2022

Characterizing the Dynamic Response of a Foam-Based Testbed with Material, Geometric, and Experimental Uncertainties

Designing Accelerated Vibration Tests using Model-Based Equivalent Damage Prediction

A Simplified Finite Element Joint Model Updated with Experimental Modal Features

Additively Manufactured Component Characterization by Machine Learning from Resonance Inspection Techniques

Increasing Multi-Axis Testing Confidence through Finite Element and Input Control Testing

Neural Radiance Fields as Digital Twins for Photorealistic Event-Based Structural Dynamics

Structural Health Monitoring in the Context of Non-Equilibrium Phase Transitions

2021

Simulating SDOF and MIMO Vibration Tests

Environmental-Insensitive Damage Features Based on Output-to-Output Coherence

Signature-Driven Adaptable Collaborative Sampling

Estimation of Light Intensity Time Series from Event-Driven Imagery for Acoustic Measurements

Waveometry: Multi-Layer Input Deep Learning Applied to Ultrasonic Wavefield Measurements

Deep Reinforcement Learning for Active Structure Stabilization

Estimation of Structural Vibration Modal Properties Using a Spike-Based Computing Paradigm

Data challenges for Structural Health Monitoring of Electrical Machines

Effective Presentations

B61-12 Life Extension Program: System Qualification and Joint Testing

An Overview of Offshore Wind Energy

Ratchet Effects and Non-Equilibrium Dynamics

An Introduction to Robotic Motion Planning

Aircraft Structural Dynamics

Experimental Modal Overview

2020

Deep Waves: Defect Characterization in Ultrasonic Wavefield Measurements using Deep Learning

Designing Operations- and Environmental-Insensitive Damage Features

Detecting Acoustic Change in a Reverberant Environment

Digital Spatio-Temporal Coded Exposures for Dynamic Characterization using Event-Driven Imagery

Hidden Layers: Detection of Composite-Metal Interface Delamination using Ultrasonic Wavefield Imaging

Informed Design of Radiation Detecting Textiles

Robust, Self-Healing Video-Based Sensor Networks

Making Effective Presentations - Rethinking Slide Design

B61-12 Life Extension Program System Qualification and Joint Testing

An Overview of Offshore Wind Energy

Verification and Validation of Computational Models

Effective Microstructure on Damage and Failure in Materials

Aircraft Structural Dynamics

LANL Satellite Engineering

2019

Applying Information Complexity Measures to Structural Health Monitoring

Assessing Predictive Capabilities for Nonlinear Dynamic Structural Responses

Characterizing the Electrical Pathways of Heart Cells from Calcium Imaging Video – An Unsupervised Machine Learning Approach

Power Sync

Real-Time Hybrid Substructuring for System Vibration Testing

Toward Extracting Multidimensional Kinematics from SWIFT Experiments

Visio-Acoustic Data Fusion

Making Effective Presentations - Rethinking Slide Design

B61-12 Life Extension Program System Qualification and Joint Testing

An Overview of Offshore Wind Energy

Verification and Validation of Computational Models

Effective Microstructure on Damage and Failure in Materials

Aircraft Structural Dynamics

LANL Satellite Engineering

2018

Imager-based Characterization of Viscoelastic Material Properties

Imager-based Techniques for Analyzing Metallic Melt Pools for Additive Manufacturing

Augmented Reality for Interactive Robot Control

Input Dimension Reduction for Materials Models

Remote Detection of Abnormal Behavior in Mechanical Systems

Wavenumber Spectroscopy Techniques for Inspection using Laser Ultrasound

Structured Dynamics and Biology