Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “time to failure”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Classification of Photovoltaic Failures with Hidden Markov Modeling, an Unsupervised Statistical Approach

Failure detection methods are of significant interest for photovoltaic (PV) site operators to help reduce gaps between expected and observed energy generation. Current approaches for field-based fault detection, however, rely on multiple data inputs and can suffer from interpretability issues. In contrast, this work offers an unsupervised statistical approach that leverages hidden Markov models (HMM) to identify failures occurring at PV sites. Using performance index data from 104 sites across the United States, individual PV-HMM models are trained and evaluated for failure detection and transition probabilities. This analysis indicates that the trained PV-HMM models have the highest probability of remaining in their current state (87.1% to 93.5%), whereas the transition probability from normal to failure (6.5%) is lower than the transition from failure to normal (12.9%) states. A comparison of these patterns using both threshold levels and operations and maintenance (O&M) tickets indicate high precision rates of PV-HMMs (median = 82.4%) across all of the sites. Although additional work is needed to assess sensitivities, the PV-HMM methodology demonstrates significant potential for real-time failure detection as well as extensions into predictive maintenance capabilities for PV.

classification↗
DOE OSTI2021

Laboratory earthquake forecasting: A machine learning competition

Earthquake prediction, the long-sought holy grail of earthquake science, continues to confound Earth scientists. Could we make advances by crowdsourcing, drawing from the vast knowledge and creativity of the machine learning (ML) community? We used Google’s ML competition platform, Kaggle, to engage the worldwide ML community with a competition to develop and improve data analysis approaches on a forecasting problem that uses laboratory earthquake data. The competitors were tasked with predicting the time remaining before the next earthquake of successive laboratory quake events, based on only a small portion of the laboratory seismic data. The more than 4,500 participating teams created and shared more than 400 computer programs in openly accessible notebooks. Complementing the now well-known features of seismic data that map to fault criticality in the laboratory, the winning teams employed unexpected strategies based on rescaling failure times as a fraction of the seismic cycle and comparing input distribution of training and testing data. In addition to yielding scientific insights into fault processes in the laboratory and their relation with the evolution of the statistical properties of the associated seismic data, the competition serves as a pedagogical tool for teaching ML in geophysics. The approach may provide a model for other competitions in geosciences or other domains of study to help engage the ML community on problems of significance.

58 GEOSCIENCES↗
DOE OSTI2022

Validation of material models for puncture of 7075-T651 aluminum plate

Plate puncture simulations are challenging computational tasks that require advanced material models including high strain rate and thermal-mechanical effects on both deformation and failure, plus finite element techniques capable of representing large deformations and material failure. The focus of this work is on the material issues, which require large sets of experiments, flexible material models and challenging calibration procedures. In this study, we consider the puncture of 12.7 mm thick, 7075-T651 aluminum alloy plates by a cylindrical punch with a hemispherical nose and diameter of 12.7 mm. The plasticity and ductile failure models were isotropic with calibration data obtained from uniaxial tension tests at different temperatures and strain rates plus quasi-static notched tension tests and shear-dominated tests described here. Sixteen puncture experiments were conducted to identify the threshold penetration energy, mode of puncture and punch acceleration during impact, The punch was mounted on a 139 kg mass and dropped on the plates with different impact speeds. Since the mass was the same in all tests, the quantity of interest was the impact speed. The axis and velocity of the punch were perpendicular to the plate surface. The mean threshold punch speed was 3.05 m/s, and the mode of failure was plugging by thermal-mechanical shear banding accompanied by scabbing fragments. Application of the material models in simulations of the tests yielded accurate estimates of the threshold puncture speed and of the mode of failure. Time histories of the punch acceleration compared well between simulation and test. Remarkably, the success of the simulations occurred in spite of even the smallest element used being larger than the width of the shear bands.

36 MATERIALS SCIENCE↗
DOE OSTI2023

Reliability modeling in a predictive maintenance context: A margin-based approach

Current system reliability methods (typically based on fault trees or reliability block diagrams) can effectively propagate reliability data from the asset to the system level in order to identify system critical points. However, employed asset reliability data are an approximated integral representation of the past industrywide operational experience, and they neglect the present asset health status (available, for example, from online monitoring data and diagnostic assessments) and forecasted health projection (when available from prognostic models). Asset health should be informed solely by that specific asset’s current and historical performance data and should not be an approximated integral representation of the past industrywide operational experience (as currently performed by system reliability models through Bayesian updating processes). Sensor data, diagnostic assessments, and prognostic assessments are in fact not considered in plant reliability models used to inform system engineers on the most critical assets. In addition, the propagation of quantitative health data from the asset to the system level is a challenge given the diverse nature and structure of health data elements (e.g., vibration spectra, temperature readings, expected failure time). Ideally, in a predictive maintenance context, system reliability models should support decision making by propagating available health information from the asset to the system level in order to provide a quantitative snapshot of system health and identify the most critical assets. Here, this paper is directly addressing these two goals by proposing a different approach for reliability modeling that relies on asset diagnostic and prognostic assessments, along with monitoring data to measure asset health. The propagation of health data from the asset to the system level is performed through fault tree models not in probability terms, but in terms of margin where margin is the “distance” between the present status and an undesired event (e.g., failure or unacceptable performance). Through a cause-effect lens, while classical reliability models target the effect associated with asset performance, a margin-based approach focuses on the cause of an undesired asset performance (i.e., its health). Hence, thinking of reliability in terms of margins implies decision-making based on causal reasoning. We will show how fault tree models can be solved using a margin language and how this process can effectively assist system engineers to identify the most critical assets.

97 - MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Profiling overstory survival trends following varying thinning and burning disturbance regimes in a mixed pine-hardwood forest in the US South

Prescribed burning as a silvicultural tool may be effective for achieving various management objectives, yet some practitioners have concerns about inadvertently increasing overstory tree mortality. Using extensive data from a long-term, ongoing study in northcentral Alabama, USA, that systematically applies a variety of thinning and prescribed burning disturbances on mixed pine-hardwood stands, we assess survival trends for different groups of trees. The primary research question is whether more frequent prescribed burns adversely affects overstory survival versus infrequent or no burns; secondary questions explore whether overstory survival trends differ based on thinning level, species group, or size class. Interest is in broad groups of trees, not individual tree survival or mortality; consequently, survival analysis methods are used, including nonparametric Kaplan-Meier (KM) techniques for examining single grouping factors, and parametric accelerated failure time models for analyzing simultaneous effects of multiple covariates. Leveraging relatively recent methodological advances, the statistical techniques are adjusted for interval censored data. KM survival curves showed that more frequent prescribed fires did not result in differing survival trends. Also, trees in unthinned stands had the lowest 14-year survival compared to thinned stands, oaks had the highest survival compared to pines and the others species group, and the smallest size class of overstory trees had the lowest survival. In conclusion, these results support managing transitional mixed pine-hardwoods using thinning and multiple prescribed fires to restore specific species composition and structure.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2021

Battery asset management with cycle life prognosis

We report Battery Asset Management problem determines the minimum cost replacement schedules for each individual asset in a group of battery assets that operate in parallel. Battery cycle life varies under different operating conditions including temperature, depth of discharge (DOD), charge rate, etc., and a battery deteriorates due to usage, which cannot be handled by current asset management models. This paper presents a new battery asset management methodology where battery cycle life prognosis is integrated with parallel asset management to reduce lifecycle cost of the Battery Energy Storage Systems (BESS). For the battery failure time prognosis, a nonlinear physics-based battery capacity fade model is developed and incorporated in parallel asset management model to update battery capacity over time. Experiment results have shown that the developed battery asset management methodology can be conveniently used to facilitate BESS asset management decision making thereby decreasing asset lifecycle costs.

25 ENERGY STORAGE↗
DOE OSTI2021

The Preservation of Climate‐Driven Landslide Dams in Western Oregon

Abstract Bedrock landsliding, including the formation of landslide dams, is a predominant geomorphic process in steep landscapes. Clarifying the importance of hydrologic and seismic mechanisms for triggering deep‐seated landslides remains an ongoing effort, and formulation of geomorphic metrics that predict dam preservation is crucial for quantifying secondary landslide hazards. Here, we identify >200 landslide‐dammed lakes in western Oregon and utilize dendrochronology and enhanced 14 C dating (“wiggle matching”) of “ghost forests” to establish slope failure timing at 20 sites. Our dated landslide dataset reveals bedrock landsliding has been common since the last Cascadia Subduction Zone earthquake in January 1700 AD. Our study does not reveal landslides that date to 1700 AD. Rather, we observe temporal clustering of at least four landslides in the winter of 1889/1890 AD, coincident with a series of atmospheric rivers that generated one of the largest regionally recorded floods. We use topographic and field analyses to assess the relation between dam preservation and topographic characteristics of the impounded valleys. In contrast to previous studies, we do not observe systematic scaling between dam size and upstream drainage area, though dam stability indices for our sites correspond with “stable” dams elsewhere. Notably, we observe that dams are preferentially preserved at drainage areas of ∼1.5 to 13 km 2 and valley widths of ∼25 to 80 m, which may reflect the reduced downstream influence of debris flows and the accumulation of mature conifer trees upstream from landslide‐dammed lake outlets. We suggest that wood accumulation upstream of landslide dams tempers large stream discharges, thus inhibiting dam incision.

Struble, William T.↗
DOE OSTI2022

Failure Probability Constrained AC Optimal Power Flow

Despite cascading failures being the central cause of blackouts in power transmission systems, existing operational and planning decisions are made largely by ignoring their underlying cascade potential. This paper posits a reliability-aware AC Optimal Power Flow formulation that seeks to design a dispatch point which has a low operator-specified likelihood of triggering a cascade starting from any single component outage. By exploiting a recently developed analytical model of the probability of component failure, our Failure Probability-constrained ACOPF (FP-ACOPF) utilizes the system's expected first failure time as a smoothly tunable and interpretable signature of cascade risk. Here, we use techniques from bilevel optimization and numerical linear algebra to efficiently formulate and solve the FP-ACOPF using off-the-shelf solvers. Extensive simulations on the IEEE 118-bus case show that, when compared to the unconstrained and N-1 security-constrained ACOPF, our probability-constrained dispatch points can significantly lower the probabilities of long severe cascades and of large demand losses, while incurring only minor increases in total generation costs.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2021

Testing in Support of the Development of EPP Plus SMT Design Method at ORNL (FY2021)

Experiments in support of the development of the integrated Elastic–Perfectly Plastic (EPP) plus Simplified Model Test (SMT) design methodology, referred to as the EPP+SMT method, continued in FY21. This report focuses on the methodology for developing the EPP+SMT creep-fatigue (CF) design curves at low strain ranges. The creep damage-based method and dissipated work-based method were used to evaluate the available CF data at the low strain range region. A set of failure criteria were determined, and a simple extrapolation method was developed to predict the CF life cycles at low strain ranges that are not accessible by experiments due to the extraordinarily long failure times at the low strain region (thousands to hundreds of thousands of years) and the inability of the test machine to control these small strain ranges due to the signal to noise issues. An experimental method with the concept of block-strain range CF testing was proposed to generate the information needed to extrapolate the CF design curves to low strain ranges. Based on this new testing approach, a preliminary EPP+SMT CF design curve was developed for Alloy 617 at 950°C with tension hold time of 100 s. The analysis in this report shows the potential of generating a set of EPP+SMT CF design curves with different hold times within a reasonable amount of time and testing effort. Based on such a progress, a hold time extrapolation procedure for low strain ranges will be developed in FY22 and critical testing will be carried out to complete the development of the EPP+SMT CF design curves for Alloy 617.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Extinction Dynamics of the Foot-and-Mouth Disease Virus Carrier State Under Natural Conditions

Foot-and-mouth disease (FMD) is one of the most economically important livestock diseases worldwide. Following the clinical phase of FMD, a large proportion of ruminants remain persistently infected for extended periods. Although extinction of this carrier state occurs continuously at the animal and population levels, studies vary widely in their estimates of the duration of persistent infection. There is a need for robust statistical models to capture the dynamics of persistent infection for the sake of guiding FMD control and trade policies. The goal of the current study was to develop and assess statistical models to describe the extinction of FMD virus (FMDV) persistent infection using data from primary longitudinal studies of naturally infected cattle and Asian buffalo in Vietnam and India. Specifically, accelerated failure time (AFT) models and generalized linear mixed models (GLMM) were developed to predict the probability of persistent infection in seropositive animals and identified carriers at the individual animal level at sequential time points after outbreaks. The primary studies were analyzed by country and combined using an individual-participant data meta-analysis approach. The models estimated similar trends in the duration of persistent infection for the study/species groups included in the analyses, however the significance of the trends differed between the models. The overall probabilities of persistent infection were similar as predicted by the AFT and GLMM models: 6 months: 99% (AFT) /80% (GLMM), 12 months: 51% (AFT) /32% (GLMM), 18 months: 6% (AFT) /5% (GLMM), 24 months: 0.8% (AFT) /0.6% (GLMM). These models utilizing diverse and robust data sets predict higher probabilities of persistence than previously published, suggesting greater endurance of carriers subsequent to an outbreak. This study demonstrates the utility of statistical models to investigate the dynamics of persistent infection and the importance of large datasets, which can be achieved by combining data from several smaller studies in meta-analyses. Results of this study enhance current knowledge of the FMDV carrier state and may inform policy decisions regarding FMDV persistent infection.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2021

Reliability metrics and their management implications for open pond algae cultivation

The prevalence of contaminating organisms in outdoor algae cultivation, with the often-associated dramatic crop failures, necessitates the need for metrics describing production system reliability. Standard metrics for algae cultivation reliability are critically needed to be able to map improvements in operational parameters, but do not currently exist. In this work, we present a set of standard metrics including mean time to failure (MTTF) and mean time between failures (MTBF) as the basis for the calculation of pond failure rate (FR) and reliability coefficient (RC). Metrics associated with the numbers of contaminating organisms such as abundance ratio (AR), prevalence of infection (PI), and mean intensity of infection (MII) are also relevant. These metrics, based on measured experimental values of outdoor pond performance during the Algae Testbed Public-Private Partnership (ATP3) Unified Field Studies (UFS), provide a basis for quantifying pond failure and provide insight into potential pond management and contaminant mitigation strategies. From these reliability metrics applied to this dataset, we are able to link operational parameters, such as harvest frequency and inoculum source, to pond reliability for different algae strains and seasons, and from an assessment of AR, provide contamination thresholds beyond which a culture may be unrecoverable. Ultimately, the implementation of widely-practiced and simple-to-calculate algae pond reliability metrics calculated from rapid and easy to collect data or observations will reduce risk and uncertainty in large-scale algae deployment and aid in the development of integrated pest management (IPM) strategies.

09 BIOMASS FUELS↗
DOE OSTI2020

Joint Development of SAS4A Code in Application to Oxide-fueled LFR Severe Accident Analysis

The scope of this project was to pursue specific SAS4A liquid-metal cooled reactor (LMR) safety analysis software extensions to simulate postulated accidents with fuel failures for oxide-fueled Lead-cooled Fast Reactors (LFRs). Since most U.S. LMR experience is on sodium-cooled fast reactor options based on past testing and operation experience with EBR-II and FFTF, the DOE’s legacy fast reactor safety analysis capabilities were focused on metal-fueled pool-type concepts with sodium coolant. In recent years, Westinghouse Electric Company (WEC) has decided to pursue an LFR design as one of their next generation nuclear technology options because of its favorable safety and economics attributes. Oxide fuel is considered among other fuel options due to previous WEC experience with this fuel form. Development of this new technology requires the availability of adequately accurate computational tools, some of which can be adapted from versions of similar software used for analysis of other LMRs. Although Argonne National Laboratory’s (ANL) SAS4A/SASSYS-1 safety analysis software suite (shortened as SAS4A code hereafter for brevity) has the basic capabilities to model LFR system designs, the SAS4A code modules used in the analysis of accidents with fuel/cladding failures lack appropriate models for the unique phenomena that govern as-irradiated oxide-fuel damage mechanisms in lead coolant. Therefore, the objective of this project was to extend the capabilities of SAS4A with mechanistic oxide-fuel failure models in lead coolant for margin to failure assessments, analysis of failure modes, location and timing of failures under different accident scenarios consistent with the whole-plant dynamic response including the reactivity feedback, and assessment of the potential for fuel damage propagation due to potential fission gas jet and fuel-fragment/molten fuel impingement to neighboring fuel pins in an assembly. This report provides mainly a summary of Argonne’s technical contributions in the joint project, but the reports and publications by the Participant team are included as references at the end of the report.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2020

Kinetics of Failure in an Elastic Peridynamic Material

The dynamic behavior of an elastic peridynamic material with a nonconvex bond potential is studied. In spite of the material’s inherently unstable nature, initial value problems can be solved using essentially the same techniques as with conventional materials, both analytically and numerically. In a suitably constructed material model, small perturbations grow exponentially over time until the material fails. Additionally, the time for this growth is computed explicitly for a stretching bar that passes from the stable to the unstable phase of the material model. This time to failure represents an incubation time for the nucleation of a crack. The finiteness of the failure time in effect creates a rate dependence in the failure properties of the material. Thus, the unstable nature of the elastic material leads to a rate effect even though it does not contain any terms that explicitly include a strain rate dependence.

36 MATERIALS SCIENCE↗
DOE OSTI2023

The effect of current and lambda on white-etch-crack failures

White etching cracks (WECs) have been associated with premature failure of wind turbine roller bearings. Various drivers for the generation of WECs have been identified such as loading conditions, slip, steel quality, lubrication, hydrogen embrittlement, corrosion fatigue cracking, and stray electrical currents passing through the surface. Here, in this work, a benchtop test rig utilizing a three-ring-on-roller test configuration was used to investigate the effect of electrical current and operation in different lubricating regimes, defined by lambda (λ), on high-quality bearing steel samples tested in a commercially available power transmission EP gear lubricant. It was observed that there is an inverse correlation between the magnitude of electric current applied to the ring/ roller system and time-to-failure. Higher current magnitudes lead to shorter time-to-failure than lower current magnitudes, with macropitting as the main failure mode. Sub-surface investigation revealed the presence of WECs in all cases. For the same current magnitude, tests conducted in boundary and mixed lubrication regimes showed that time-to-failure increased as lambda increased, and the tests resulted in WEC related macropits, whereas tests conducted in near-hydrodynamic regime resulted in surface damage with no macropit. It was also noted that a shift toward near-hydrodynamic lubrication resulted in a distinct surface distress on the roller surface. Furthermore, there seems to be a transition in the mixed regime during which the surface distress occurred. The damage on the surface of the test samples resembled non-spatially, periodic, groove-like corrugations and, in some cases, crater-like depressions. Sub-surface imaging, performed by sequential sectioning, revealed the presence of WECs in all cases, and broad, branching cracks that were more prevalent under the more severe boundary conditions.

17 WIND ENERGY↗
DOE OSTI2021

Summary of Previous Mechanical Test Data on ODS Alloys 14YWT and OFRAC up to 1000ºC

The Nanostructured Ferritic Alloys (NFA) 14YWT and OFRAC were developed for future fission and fusion nuclear energy reactors requiring high-temperature mechanical properties that are tolerant to extreme neutron irradiation environments. The NFA contain a high concentration of Ti-, Y- and O-enriched nanoclusters (NC) and ultra-fine grains to achieve high temperature strength and creep properties and high sink strength for trapping irradiation induced point defects to minimize hardening and swelling and transmutated He atoms to form intragranular nano-size bubbles that prevent formation of coarse bubbles on grain boundaries that cause embrittlement. The mechanical properties of 14YWT and OFRAC have been acquired from tensile and creep tests conducted in the past at Oak Ridge National Laboratory. The development of 14YWT started in 2000, resulting in the production of numerous heats. Tensile properties were obtained from eleven heats of 14YWT from room temperature to 800ºC. Tensile data for the SM10 heat of 14YWT was extended to 1,000ºC. Initial development of OFRAC occurred in 2016. Tensile tests conducted from room temperature to 800ºC revealed similar properties of OFRAC with those obtained from the newer generation of 14YWT heats. The creep properties of 14YWT-SM10 evaluated using constant stress tensile and load time-to-failure tests at 800ºC showed low minimum creep rates with stresses of 200 and 100 MPa. The single time-to-failure test of 14YWT-SM10 at 800ºC and 100 MPa was terminated after 20,357 hours with no specimen failure and a low creep strain of ~0.24 %. The creep properties of OFRAC determined from strain-rate jump tests at were similar those of 14YWT-SM10. The stress exponent for 14YWT and OFRAC at 800ºC are similar and are consistent with threshold stress behavior. Since both 14YWT and OFRAC are candidates for fuel cladding in future fast reactors, several fabrication studies were recently conducted and have successfully demonstrated that thin wall tubes can be fabricated from 14YWT and OFRAC by cold pilger rolling and high precision tube rolling. The high temperature mechanical properties and feasibility of fabricating thin wall tubes make 14YWT and OFRAC candidates for application as heat pipes in advanced micro-reactors. The purpose of this report is to summarize the previously obtained tensile and creep data at temperatures up to 1000ºC for NFA 14YWT and OFRAC that have been acquired over the past 20 years at ORNL.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2021

Resiliency in numerical algorithm design for extreme scale simulations

Here this work is based on the seminar titled ‘Resiliency in Numerical Algorithm Design for Extreme Scale Simulations’ held March 1–6, 2020, at Schloss Dagstuhl, that was attended by all the authors. Advanced supercomputing is characterized by very high computation speeds at the cost of involving an enormous amount of resources and costs. A typical large-scale computation running for 48 h on a system consuming 20 MW, as predicted for exascale systems, would consume a million kWh, corresponding to about 100k Euro in energy cost for executing 10 23 floating-point operations. It is clearly unacceptable to lose the whole computation if any of the several million parallel processes fails during the execution. Moreover, if a single operation suffers from a bit-flip error, should the whole computation be declared invalid? What about the notion of reproducibility itself: should this core paradigm of science be revised and refined for results that are obtained by large-scale simulation? Naive versions of conventional resilience techniques will not scale to the exascale regime: with a main memory footprint of tens of Petabytes, synchronously writing checkpoint data all the way to background storage at frequent intervals will create intolerable overheads in runtime and energy consumption. Forecasts show that the mean time between failures could be lower than the time to recover from such a checkpoint, so that large calculations at scale might not make any progress if robust alternatives are not investigated. More advanced resilience techniques must be devised. The key may lie in exploiting both advanced system features as well as specific application knowledge. Research will face two essential questions: (1) what are the reliability requirements for a particular computation and (2) how do we best design the algorithms and software to meet these requirements? While the analysis of use cases can help understand the particular reliability requirements, the construction of remedies is currently wide open. One avenue would be to refine and improve on system- or application-level checkpointing and rollback strategies in the case an error is detected. Developers might use fault notification interfaces and flexible runtime systems to respond to node failures in an application-dependent fashion. Novel numerical algorithms or more stochastic computational approaches may be required to meet accuracy requirements in the face of undetectable soft errors. These ideas constituted an essential topic of the seminar. The goal of this Dagstuhl Seminar was to bring together a diverse group of scientists with expertise in exascale computing to discuss novel ways to make applications resilient against detected and undetected faults. In particular, participants explored the role that algorithms and applications play in the holistic approach needed to tackle this challenge. This article gathers a broad range of perspectives on the role of algorithms, applications and systems in achieving resilience for extreme scale simulations. The ultimate goal is to spark novel ideas and encourage the development of concrete solutions for achieving such resilience holistically.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

OLCF Summit Supercomputer GPU Snapshots During Double-Bit Errors and Normal Operations

As we move into the exascale era, the power and energy footprints of high-performance computing (HPC) systems have grown significantly larger. Due to the harsh power and thermal conditions the system, components are exposed to extreme operating conditions. Operation of such modern HPC systems requires deep insights into long term system behavior to maintain its efficiency as well as its longevity. To help the HPC community to gain such insights, we provide double-bit errors using system telemetry data and logs collected from the Summit supercomputer, equipped with 27,648 Tesla V100 GPUs with 2nd-generation high-bandwidth memory (HBM2). The dataset relies on Nvidia XID records internally collected by GPU firmware at the time of failure occurrence, on the reboot-time logs of each Summit node, on node-level job scheduler records collected after each job termination, and on a 1Hz data rate from the baseboard management controllers (BMCs) of each Summit compute node using the OpenBMC event subscription protocol. Technical details can be found in the paper Oles et. al “Understanding GPU Memory Corruption at Extreme Scale: The Summit Case Study” ICS’24 (https://doi.org/10.1145/3650200.3656615).

97 MATHEMATICS AND COMPUTING↗