Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “failure prediction”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Failure prediction and estimation of failure parameters

Machine-learning methods and apparatus are disclosed to determine frictional state or other parameters in an earthquake zone or other failing medium, using acoustic emission, seismic waves, or other detectable indicators of microscopic processes. Predictions of future failures are demonstrated in different regimes. A classifier is trained using time series of acoustic emission data along with historic data of frictional state or failure events. In disclosed examples, random forests and gradient boost trees are used, and grid-search or EGO procedures are used for hyperparameter tuning. Once trained, the classifier can be applied to testing or live data in order to assess a frictional state, assess seismic hazard, or make predictions regarding a future failure event. The technology has been developed in a double direct shear apparatus, but can be widely applied to seismic faults, other terrestrial failures, or failures in man-made structures. Variations are disclosed.

Johnson, Paul Allan↗
DOE OSTI2022

Predicting failure parameters of semiconductor devices subjected to stress conditions

A method for predicting failure parameters of semiconductor devices can include receiving a set of data that includes (i) characteristics of a sample semiconductor device, and (ii) parameters characterizing a stress condition. The method further includes extracting a plurality of feature values from the set of data and inputting the plurality of feature values into a trained model executing on the one or more processors, wherein the trained model is configured according to an artificial intelligence (AI) algorithm based on a previous plurality of feature values, and wherein the trained model is operable to output a failure prediction based on the plurality of feature values. Further, the method includes generating, via the trained model, a predicted failure parameter of the sample semiconductor device due to the stress condition.

Ahmed, Moinuddin↗
DOE OSTI2023

Predicting Failure Using Deep Learning SAND Report

Accurate prediction of ductile failure is critical to Sandia’s NW mission, but the models are computationally heavy. The costs of including high-fidelity physics and mechanics that are germane to the failure mechanisms are often too burdensome for analysts either because of the person-hours it requires to input them or because of the additional computational time, or both. In an effort to deliver analysts a tool for representing these phenomena with minimal impact to their existing workflow, our project sought to develop modern data-driven methods that would add microstructural information to business-as-usual calculations and expedite failure predictions. The goal is a tool that receives as input a structural model with stress and strain fields, as well as a machine-learned model, and output predictions of structural response in time, including failure. As such, our project spent substantial time performing high-fidelity, three-dimensional experiments to elucidate materials mechanisms of void nucleation and evolution. We developed crystal-plasticity finite-element models from the experimental observations to enrich the findings with fields not readily measured. We developed engineering length-scale simulations of replicated test specimens to understand how the engineering fields evolve in the presence of fine-scale defects. Finally, we developed deep learning convolutional neural networks, and graph-based neural networks to encode the findings of the experiments and simulations and make forward predictions in time for structural performance. This project demonstrated the power of data-driven methods for model development, which have the potential to vastly increase both the accuracy and speed of failure predictions. These benefits and the methods necessary to develop them are highlighted in this report. However, many challenges remain to implementing these in real applications, and these are discussed along with potential methods for overcoming them.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Modeling strategy for progressive failure prediction in lithium-ion batteries under mechanical abuse

The prediction for the internal failure of lithium-ion batteries (LIBs) under external mechanical abuse loading remains a challenge for safe design. This paper systematically studies the modeling approach for progressive failure simulation and short-circuit prediction. Hemispherical indentation experiments are conducted on a battery specimen containing 5 representative sandwich (RS) layers to examine the deformation and progressive failure behavior of battery components. Three different models, viz, a high-fidelity detailed model, an intermediate homogenized model and a fully homogenized model, are developed and employed to simulate the progressive failure in a multi-layer battery specimen. The simulation results obtained using the three models all correlate well with experimental phenomena, capturing the overall stress-strain response and ultimate failure of the battery structure. The numerical results can help us understand the failure behavior and the mechanism for mechanical abuse-induced short circuit in LIBs. The feasibility of the three models for use in progressive failure prediction is compared by analyzing the capability of the models in predicting the critical failure events, the computational efficiency of the models, and the feasibility to implement coupled mechanical-electrochemical-thermal simulations. The results of this work provide useful insights on practical choices for the modeling strategy and safety design of LIBs under mechanical abuse conditions.

25 ENERGY STORAGE↗
DOE OSTI2021

Simulation of hardened cement degradation and estimation of uncertainty in predicted failure times with peridynamics

Modeling the degradation of cement-based infrastructure due to aqueous environmental conditions continues to be a challenge. In order to develop a capability to predict concrete infrastructure failure due to chemical degradation, here we created a chemomechanical model of the effects of long-term water exposure on cement paste. The model couples the mechanical static equilibrium balance with reactive–diffusive transport and incorporates fracture and failure via peridynamics (a meshless simulation method). The model includes fundamental aspects of degradation of ordinary Portland cement (OPC) paste, including the observed softening, reduced toughness, and shrinkage of the cement paste, and increased reactivity and transport with water induced degradation. This version of the model focuses on the first stage of cement paste decalcification, the dissolution of portlandite. Given unknowns in the cement paste degradation process and the cost of uncertainty quantification (UQ), we adopt a minimally complex model in two dimensions (2D) in order to perform sensitivity analysis and UQ. We calibrate the model to existing experimental data using simulations of common tests such as flexure, compression and diffusion. Then we calculate the global sensitivity and uncertainty of predicted failure times based on variation of eleven unique and fundamental material properties. We observed particularly strong sensitivities to the diffusion coefficient, the reaction rate, and the shrinkage with degradation. Also, the predicted time of first fracture is highly correlated with the time to total failure in compression, which implies fracture can indicate impending degradation induced failure; however, the distributions of the two events overlap so the lead time may be minimal. Extension of the model to include the multiple reactions that describe complete degradation, viscous relaxation, post-peak load mechanisms, and to three dimensions to explore the interactions of complex fracture patterns evoked by more realistic geometry is straightforward and ongoing.

36 MATERIALS SCIENCE↗
DOE OSTI2021

A method for predicting failure statistics for steady state elevated temperature structural components

This paper presents the initial development of a high temperature life prediction method that accounts for the variability in the material properties of Grade 91 steel. The method accounts for material variability by fitting a variable 3-parameter Weibull distribution to experimental rupture data and accounts for the variability of creep deformation on the steady-state stresses via a Monte Carlo approach. To ensure reasonable computational times, the model represents the material as an extremely viscous Stokes fluid with a non-Newtonian viscosity, therefore solving the stress relaxation problem with a steady, static, instead of transient, analysis. Furthermore, the complete statistical analysis combines this model for creep deformation with a probabilistic model for creep rupture to evaluate the probability of premature failure for a set of sample problems, comparing the predicted failure statistics to the design life predicted by the ASME Boiler and Pressure Vessel Code rules.

42 ENGINEERING↗
DOE OSTI2020

Mesh Failure Prediction Using Deep Learning Techniques

The goal of this joint effort between SRI and LLNL has been to investigate the use of state-of-the-art deep learning architectures to effectively explore the correlations between a set of mesh zone attributes and manifestations of failure in a given mesh simulation and draw useful conclusions towards a practical failure detection system. Each zone attribute is a scalar feature that is recorded separately for each zone during simulation, resulting in a multivariate time series signal for each zone. Currently, these features are used to manually decide where to apply manual relaxations so that simulations run without any failures. Our long-term vision is to build a machine learning system that is capable of reliably detecting upcoming failures and is able to apply finer-grained targeted relaxations automatically to relevant zones to avoid an expected failure.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

A liquid stratification model to predict failure in thermally damaged EBW detonators

In previous work, commercially available downward facing exploding bridgewire detonators (EBWs) were exposed to elevated temperatures. These detonators were then initiated using a firing set which discharged a high amplitude short duration electrical pulse into a thin gold bridgewire. Responses of the detonators were measured using photonic doppler velocimetry (PDV) and high-speed photography. A time delay of 4 μs between EBW initiation and first movement of an output flyer separated operable detonators from inoperable detonators or duds. Here, we propose a simple method to determine detonator operability from the calculated state of the detonator at the time the firing set is initiated. The failure criterion is based on the gap distance between the exploding bridgewire (EBW) and the adjacent initiating explosive within the detonator which is low-density pentaerythritol tetranitrate (PETN) that melts between 413-415 K (140-142 ºC). The gap forms as PETN melts and flows to the bottom of the input pellet. Melting of PETN is modeled thermodynamically as an energy sink using a normal distribution spread over a temperature range between the onset temperature of 413 K and the ending temperature of 415 K. The extent of the melt is determined from the average temperature of the PETN. The PETN liquid is assumed to occupy the interstitial gas volume in the lower part of the input pellet. The vacated volume from the relocated liquid forms the gap between the EBW and the PETN. The remaining sandwiched layer consists of solid PETN particles and gas filling interstitial volume. We predict that a threshold gap between 17-27 μm separates properly functioning detonators from duds.

45 MILITARY TECHNOLOGY, WEAPONRY, AND NATIONAL DEF↗
DOE OSTI2020

Orchestrating Fault Prediction with Live Migration and Checkpointing

Checkpoint/Restart (C/R) is widely used to provide fault tolerance on High-Performance Computing (HPC) systems. However, Parallel File System (PFS) overhead and failure uncertainty cause significant application overhead. This paper develops an adaptive multi-level C/R model that incorporates a failure prediction and analysis model, which orchestrates failure prediction, checkpointing, checkpoint frequency, and proactive live migration along with the additional benefit of Burst Buffers (BB). It effectively reduces the overheads due to failures, checkpointing, and recovery. Simulation results for the Summit supercomputer yield a reduction of ~20%-86% in application overhead due to BBs, orchestrated failure prediction, and migration. We also observe a ~29% decrease in checkpoint writes to BBs, which can increase the longevity of the BB storage devices.

Behera, Subhendu↗
DOE OSTI2025

Representing Complex Systems as Graphs for Debugging and Predictive Maintenance-Preliminary Thoughts

Representing complex systems as graphs enables use of mathematical tools to identify faults or predict failures. Graph nodes correspond to individual modules or subsystems, and edges link coupled system parts. ‘Probes’ measure the node outputs, monitoring the system health for unexpected behavior. Assuming one cannot probe every point, within a system, the fault correlates to a region—not necessarily the specific location. Bayesian networks trained to understand fault patterns can accurately identify the source. The diagnostic tool described aides debugging by pinpointing system failure causes. For predictive maintenance, probe data develop probability distribution functions describing subsystem mean time to failure. Unit lifetime can be estimated through these probability distributions. Two approaches include using Bayesian classifiers to infer the system failure source and developing maintenance schedules by treating systems as collections of random variables. When failure behavior does not follow a closed form function, use of similarity models is proposed.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Anomaly Detection in Accelerator Facilities Using Machine Learning

Synchrotron light sources are user facilities and usually run about 5000 hours per year to support many beamlines operations in parallel. Reliability is a key parameter to evaluate machine performance. Even many facilities have achieved >95% beam reliability, there are still many hours of unscheduled downtime and every hour lost is a waste of operation costs along with a big impact on individual scheduled user experiments. Preventive maintenance on subsystems and quick recovery from machine trips are the basic strategies to achieve high reliability, which heavily depends on experts’ dedication. Recently, SLAC, APS, and NSLS-II collaborated to develop machine-learning-based approaches aiming to solve both situations, hardware failure prediction and machine failure diagnosis to find the root sources. In this paper, we report our facility operation status, development progress, and plans.

Accelerator Physics↗
DOE OSTI2022

Experiments and modelling of adhesive failure initiation of an epoxy underfill: a study of electronic packaging survivability

Epoxy underfills can be implemented in electronic packaging to enhance solder joint reliability of surface mounted components. However, it is important for an engineer to have a failure criterion that can be used for failure predictions and redesign of electronic assemblies. For this study, data from epoxy bond failure in mock electronic part assemblies were correlated to finite element analyses to predict adhesive failure initiation. Experiments were performed to determine failure loads for various loading locations and nonlinear viscoelastic analyses were performed for the same loading locations to determine a maximum principal strain failure parameter. Predictions showed that a maximum principal strain failure parameter defined from one test could be used as an indicator of adhesive failure of an epoxy bond undergoing other modes of loading. Failure initiation predictions matched experimental data using a maximum principal strain failure parameter for an epoxy bond undergoing mixed modes of loading for both unfilled and alumina oxide filled 828DEA epoxy. Such experimental setup is deemed appropriate for future epoxy testing.

42 ENGINEERING↗
DOE OSTI2023

Using a physics-informed neural network and fault zone acoustic monitoring to predict lab earthquakes

Abstract Predicting failure in solids has broad applications including earthquake prediction which remains an unattainable goal. However, recent machine learning work shows that laboratory earthquakes can be predicted using micro-failure events and temporal evolution of fault zone elastic properties. Remarkably, these results come from purely data-driven models trained with large datasets. Such data are equivalent to centuries of fault motion rendering application to tectonic faulting unclear. In addition, the underlying physics of such predictions is poorly understood. Here, we address scalability using a novel Physics-Informed Neural Network (PINN). Our model encodes fault physics in the deep learning loss function using time-lapse ultrasonic data. PINN models outperform data-driven models and significantly improve transfer learning for small training datasets and conditions outside those used in training. Our work suggests that PINN offers a promising path for machine learning-based failure prediction and, ultimately for improving our understanding of earthquake physics and prediction.

42 ENGINEERING↗