Search NASA⌕ Search

SEARCH · Search NASA

Results for “Faults”

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 343 records · Page 19
All sourcesNASA NTRSDOE OSTI
NASA NTRS2006

Methodology for Designing Fault-Protection Software

A document describes a methodology for designing fault-protection (FP) software for autonomous spacecraft. The methodology embodies and extends established engineering practices in the technical discipline of Fault Detection, Diagnosis, Mitigation, and Recovery; and has been successfully implemented in the Deep Impact Spacecraft, a NASA Discovery mission. Based on established concepts of Fault Monitors and Responses, this FP methodology extends the notion of Opinion, Symptom, Alarm (aka Fault), and Response with numerous new notions, sub-notions, software constructs, and logic and timing gates. For example, Monitor generates a RawOpinion, which graduates into Opinion, categorized into no-opinion, acceptable, or unacceptable opinion. RaiseSymptom, ForceSymptom, and ClearSymptom govern the establishment and then mapping to an Alarm (aka Fault). Local Response is distinguished from FP System Response. A 1-to-n and n-to- 1 mapping is established among Monitors, Symptoms, and Responses. Responses are categorized by device versus by function. Responses operate in tiers, where the early tiers attempt to resolve the Fault in a localized step-by-step fashion, relegating more system-level response to later tier(s). Recovery actions are gated by epoch recovery timing, enabling strategy, urgency, MaxRetry gate, hardware availability, hazardous versus ordinary fault, and many other priority gates. This methodology is systematic, logical, and uses multiple linked tables, parameter files, and recovery command sequences. The credibility of the FP design is proven via a fault-tree analysis "top-down" approach, and a functional fault-mode-effects-and-analysis via "bottoms-up" approach. Via this process, the mitigation and recovery strategy(s) per Fault Containment Region scope (width versus depth) the FP architecture.

Barltrop, Kevin↗
NASA NTRS2012

Fault Tolerance Middleware for a Multi-Core System

Fault Tolerance Middleware (FTM) provides a framework to run on a dedicated core of a multi-core system and handles detection of single-event upsets (SEUs), and the responses to those SEUs, occurring in an application running on multiple cores of the processor. This software was written expressly for a multi-core system and can support different kinds of fault strategies, such as introspection, algorithm-based fault tolerance (ABFT), and triple modular redundancy (TMR). It focuses on providing fault tolerance for the application code, and represents the first step in a plan to eventually include fault tolerance in message passing and the FTM itself. In the multi-core system, the FTM resides on a single, dedicated core, separate from the cores used by the application. This is done in order to isolate the FTM from application faults and to allow it to swap out any application core for a substitute. The structure of the FTM consists of an interface to a fault tolerant strategy module, a responder module, a fault manager module, an error factory, and an error mapper that determines the severity of the error. In the present reference implementation, the only fault tolerant strategy implemented is introspection. The introspection code waits for an application node to send an error notification to it. It then uses the error factory to create an error object, and at this time, a severity level is assigned to the error. The introspection code uses its built-in knowledge base to generate a recommended response to the error. Responses might include ignoring the error, logging it, rolling back the application to a previously saved checkpoint, swapping in a new node to replace a bad one, or restarting the application. The original error and recommended response are passed to the top-level fault manager module, which invokes the response. The responder module also notifies the introspection module of the generated response. This provides additional information to the introspection module that it can use in generating its next response. For example, if the responder triggers an application rollback and errors are still occurring, the introspection module may decide to recommend an application restart.

Some, Raphael R.↗
NASA NTRS2013

Validation of Helicopter Gear Condition Indicators Using Seeded Fault Tests

A "seeded fault test" in support of a rotorcraft condition based maintenance program (CBM), is an experiment in which a component is tested with a known fault while health monitoring data is collected. These tests are performed at operating conditions comparable to operating conditions the component would be exposed to while installed on the aircraft. Performance of seeded fault tests is one method used to provide evidence that a Health Usage Monitoring System (HUMS) can replace current maintenance practices required for aircraft airworthiness. Actual in-service experience of the HUMS detecting a component fault is another validation method. This paper will discuss a hybrid validation approach that combines in service-data with seeded fault tests. For this approach, existing in-service HUMS flight data from a naturally occurring component fault will be used to define a component seeded fault test. An example, using spiral bevel gears as the targeted component, will be presented. Since the U.S. Army has begun to develop standards for using seeded fault tests for HUMS validation, the hybrid approach will be mapped to the steps defined within their Aeronautical Design Standard Handbook for CBM. This paper will step through their defined processes, and identify additional steps that may be required when using component test rig fault tests to demonstrate helicopter CI performance. The discussion within this paper will provide the reader with a better appreciation for the challenges faced when defining a seeded fault test for HUMS validation.

Dempsey, Paula↗
DOE OSTI2014

ANALYSIS OF FAULT SEAL POTENTIAL FOR KNOX RESERVOIRS IN THE SOUTHERN ILLINOIS BASIN

The presence of known faults near potential geologic CO2 sequestration sites significantly raises the uncertainty of having a sufficient seal to prevent leakage along the fault plane from the intended reservoir. In regions where relocating a large sequestration project a considerable distance away from any known faults is impractical, a detailed analysis of the sealing potential of any faults within the projected future injection plume must be performed. In order to estimate the sealing potential of faults within the Late Cambrian-Early Ordovician Knox Supergroup in the Illinois Basin, two well-based cross sections were produced across two different regional fault systems (Rough Creek Fault Zone in Kentucky, and the unnamed core fault of the LaSalle Anticlinorium in Illinois) to calculate subsurface stratigraphic juxtapositions across each fault zone. Using this stratigraphic and lithologic data, three different algorithms were used to calculate the sealing potential of a theoretical Knox reservoir at each section location. These results indicate a high probability for sealing within the Rough Creek Fault Zone, but a much lower probability for a continuous seal within the LaSalle Anticlinorium.

Hickman, John↗
DOE OSTI2022

Improved Control Strategy of Grid-Forming Inverters for Fault Ride-Through in a Microgrid System

This paper develops an improved control strategy of grid-forming (GFM) inverters with fault ride-through capabilities to guarantee the stable operation of microgrids under fault conditions, especially islanded microgrids and asymmetrical faults. The proposed control strategy includes the dual control of positive-sequence and negative-sequence control as well as the adaptive virtual impedance (VI) control. Unlike existing works, the proposed strategy applies the VI control for only the d component of the positive-sequence control and leaves the q component of the positive-sequence control and the dq component of the negative-sequence control as zero, thus achieving improved stability and balanced three-phase voltages under asymmetrical faults. The adaptive feature of the VI control guarantees the stability of the GFM inverter under severe faults, which could cause the saturation of the inner current loop and instability if the VI is not adaptive. Simulation results of various unbalanced faults with high- and low-fault impedances show that the proposed control strategy improves the stability of the GFM inverter and achieves stable and balanced output voltages in islanded microgrids. And the algorithm also improves the stability of GFM inverters under balanced faults with high- and low-fault impedances.

adaptive virtual impedance control↗
DOE OSTI2012

Solid-State Fault Current Limiter Development: Design and Testing Update of a 15kV SSCL Power Stack

The Solid-State Fault Current Limiter (SSCL) is a promising technology that can be applied to utility power delivery systems to address the problem of increasing fault currents associated with load growth. As demand continues to grow, more power is added to utility system either by increasing generator capacity or by adding distributed generators, resulting in higher available fault currents, often beyond the capabilities of the present infrastructure. The SSCL is power-electronics based equipment designed to work with the present utility system to address this problem. The SSCL monitors the line current and dynamically inserts additional impedance into the line in the event of a fault being detected. The SSCL is based on a modular design and can be configured for 5kV through 69kV systems at nominal current ratings of 1000A to 4000A. Results and Findings This report provides the final test results on the development of 15kV class SSCL single phase power stack. The scope of work included the design of the modular standard building block sub-assemblies, the design and manufacture of the power stack and the testing of the power stack for the key functional tests of continuous current capability and fault current limiting action. Challenges and Objectives Solid-State Current Limiter technology impacts a wide spectrum of utility engineering and operating personnel. It addresses the problems associated with load growth both at Transmission and Distribution class networks. The design concept is pioneering in terms of developing the most efficient and compact power electronics equipment for utility use. The initial test results of the standard building blocks are promising. The independent laboratory tests of the power stack are promising. However the complete 3 phase system needs rigorous testing for performance and reliability. Applications, Values, and Use The SSCL is an intelligent power-electronics device which is modular in design and can provide current limiting or current interrupting capabilities. It can be applied to variety of applications from distribution class to transmission class power delivery grids and networks. It can also be applied to single major commercial and industrial loads and distributed generator supplies. The active switching of devices can be further utilized for protection of substation transformers. The stress on the system can be reduced substantially improving the life of the power system. It minimizes the voltage sag by speedy elimination of heavy fault currents and promises to be an important element of the utility power system. DOE Perspective This development effort is now focused on a 15kV system. This project will help mitigate the challenges of increasing available fault current. DOE has made a major contribution in providing a cost effective SSCL designed to integrate seamlessly into the Transmission and Distribution networks of today and the future. Approach SSCL development program for a 69kV SSCL was initiated which included the use of the Super GTO advanced semiconductor device which won the 2007 R&D100 Award. In the beginning, steps were identified to accomplish the economically viable design of a 69kV class Solid State Current Limiter that is extremely reliable, cost effective, and compact enough to be applied in urban transmission. The prime thrust in design and development was to encompass the 1000A and the 3000A ratings and provide a modular design to cover the wide range of applications. The focus of the project was then shifted to a 15kV class SSCL. The specifications for the 15kV power stack are reviewed. The design changes integrated into the 15kV power stack are discussed. In this Technical Update the complete project is summarized followed by a detailed test report. The power stack independent high voltage laboratory test requirements and results are presented. Keywords Solid State Current Limiter, SSCL, Fault Current Limiter, Fault Current Controller, Power electronics controller, Intelligent power-electronics Device, IED

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2023

Air Conditioning Systems Fault Detection and Diagnosis-Based Sensing and Data-Driven Approaches

The air conditioning (AC) system is the primary building end-use contributor to the peak demand for energy. The energy consumed by this system has grown as fast as it has in the last few decades, not only in the residential section but also in the industry and transport sectors. Therefore, to combat energy crises, urgent actions on energy efficiency should be taken to support energy security. Consequently, the faults in AC system components increase energy consumption due to the degradation of the system’s performance and the losses in the energy conversion procedure. In this work, AC system fault detection and diagnosis (FDD) methods are investigated to propose analytic tools to identify faults and provide solutions to those problems. The analysis of existing work shows that data-driven approaches are more accurate for both soft and hard fault detection and diagnosis in AC systems. Therefore, the proposed methods are not accurate for simultaneous fault detection, while in some works, authors tested the method with several faults separately without investigating scenarios that combine more than one fault. Moreover, this study shows that integrating data-driven approaches requires deploying an optimal sensing and measurement architecture that can detect a maximum number of faults with minimally deployed sensors. The new sensing, information, and communication technologies are discussed for their integration in AC system monitoring in order to optimize system operation and detect faults.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2022

Demonstration of a fault impact reduction control module for wind turbines

Abstract. Traditionally, wind turbines in distributed applications make control decisions as isolated systems. They generally provide maximum power output during operation and manage internal faults with little consideration of the rest of the power system. Although fault detection and tolerance schemes are widely researched and implemented, controls to ameliorate such faults are uncommon in research and industry. The rapid shutdown of a wind turbine in a large transmission-connected wind plant will have a minimal impact on a large power system, but in a microgrid or isolated grid context the abrupt loss of a single wind turbine may cause grid instability and high stress on the system. This paper demonstrates a fault impact reduction control (FIRC) module for a wind turbine, which implements wider warning thresholds around fault thresholds. When the turbine crosses a warning threshold, the controller sends its predicted action to the grid controller, which facilitates the grid operator’s response to a potential wind turbine fault and then takes appropriate action to ameliorate the fault. Various test cases demonstrate the controller action under a variety of faults, and various scenarios demonstrate the grid benefit of an FIRC in both microgrid- and grid-connected contexts. The FIRC maximizes wind turbine generation and eases generation transition under a variety of fault scenarios. The FIRC module is easy to integrate with most existing controllers, just requiring derate capability, and can be easily modified to include the various warnings and thresholds that the user desires. This analysis is mainly performed in a MATLAB-Simulink-based research wind turbine model and is also implemented in the existing LabVIEW-based controller of the same research turbine at NREL.

17 WIND ENERGY↗
DOE OSTI2022

Improved Control Strategy of Grid-Forming Inverters for Fault Ride-Through in a Microgrid System

This paper develops an improved control strategy of grid-forming (GFM) inverters with fault ride-through (FRT) capabilities to guarantee the stable operation of microgrids under fault conditions, especially islanded microgrids and asymmetrical faults. The proposed control strategy includes the dual control of positive-sequence and negative-sequence control as well as the adaptive virtual impedance (VI) control. Unlike existing works, the proposed strategy applies the VI control for only the d component of the positive-sequence control and leaves the q component of the positive-sequence control and the dq component of the negative-sequence control as zero, thus achieving improved stability and balanced three-phase voltages under asymmetrical faults. The adaptive feature of the VI control guarantees the stability of the GFM inverter under severe faults, which could cause the saturation of the inner current loop, and the instability if the VI is not adaptive. Simulation results of various unbalanced faults with high and low fault impedances show that the proposed control strategy improves the stability of the GFM inverter and achieves stable and balanced output voltages in islanded microgrids. And the algorithm also improves the stability of GFM inverters under balanced faults with high and low fault impedances.

adaptive virtual impedance control↗
DOE OSTI2020

An experimental study of the behavior of a high efficiency residential heat pump in cooling mode with common installation faults imposed

Faults are believed to be common in split system air-conditioner and heat pump systems, due to operating degradation or installation problems. Common faults include improper refrigerant charge, reduced evaporator airflow, liquid line restrictions, and the presence of non-condensable gas. Laboratory tests were used to quantify fault impacts on performance of a heat pump system, and impacts on indicator variables such as refrigerant temperatures and pressures. This paper describes a large set of laboratory tests implemented on a high efficiency heat pump operating in cooling mode. This system uses R-410A refrigerant and has a rotary compressor, TXV, two accumulators, and a compensator. A small number of previous experimental studies have been done previously to study the effects of some operating faults, but none has examined a modern system with these components. The tests were conducted with a range of fault intensities and driving conditions. The results are compared to previous researchers’ experimental results. The system’s performance was found to be quite robust in the presence of faults. The TXV, accumulators, and compensator significantly reduce sensitivity to refrigerant charge and liquid line restriction faults, and other faults to a lesser extent.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2020

Impact of installation faults in air conditioners and heat pumps in single-family homes on U.S. energy usage

To ensure that new residential air conditioners and heat pumps operate optimally, care must be taken during the initial system design and installation to avoid missteps that immediately degrade a system’s performance. Laboratory and simulation studies have quantified the impact that faults can have on capacity and efficiency. Field studies have demonstrated that installation-related faults are commonplace. However, the national impact of installation-related faults cannot be accurately estimated by only simulating a limited number of homes at several fault levels because of the variety of home characteristics and climates involved. In our analysis, we use an improved residential building stock simulation tool to predict the annual energy increase and additional utility costs resulting from two common installation faults: indoor airflow rate and refrigerant charge level. Our method considers the wide range of building characteristics and climate zones of the U.S. housing stock. We use existing field data to develop fault intensity probability distributions to inform our analysis. The study shows that these two faults result in approximately 20.7 TWh/y of additional energy use for central air conditioners and air-source heat pumps in U.S. single-family detached homes, which is a 9% increase over baseline (no-fault) usage, costing homeowners approximately $2.5 billion annually on utility bills. Air-source heat pumps are responsible for a disproportionate fraction of this energy use increase because of the larger number of operating hours compared to central air conditioners and the sensitivity of heating mode performance to the faults analyzed compared to cooling mode.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2017

Analysis of a Complex Faulted CO 2 Reservoir Using a Three-dimensional Hydro-geochemical-Mechanical Approach

This work applies a three-dimensional (3D) multiscale approach recently developed to analyze a complex CO 2 faulted reservoir that includes some key geological features of the San Andreas and nearby faults. The approach couples the STOMP-CO2-R code for flow and reactive transport modeling to the ABAQUS ® finite element package for geomechanical analysis. The objective is to examine the coupled hydro-geochemical-mechanical impact on the risk of hydraulic fracture and fault slip in a complex and representative CO 2 reservoir that contains two nearly parallel faults. STOMP-CO2-R/ABAQUS ® coupled analyses of this reservoir are performed assuming extensional and compressional stress regimes to predict evolutions of fluid pressure, stress and strain distributions as well as potential fault failure and leakage of CO 2 along the fault damage zones. The tendency for the faults to slip and pressure margin to fracture are examined in terms of stress regime, mineral composition, crack distributions in the fault damage zones and geomechanical properties. Here, this model in combination with a detailed description of the faults helps assess the coupled hydro-geochemical-mechanical effect.

58 GEOSCIENCES↗
DOE OSTI2024

Emulation and detection of physical faults and cyber-attacks on building energy systems through real-time hardware-in-the-loop experiments

The increasing use of remote or mobile access, integrated wearable technologies, data exchange, and cloud-based data analytics in modern smart buildings is steering the building industry towards open communication technologies. The increased connectivity and accessibility could lead to more cyber-attacks in smart buildings. On the other hand, physical faults (e.g., HVAC -heating, ventilation, and air-conditioning faults) may have similar adverse impacts as those from the cyber-attacks on building energy systems, such as occupant discomfort, energy wastage, and equipment downtime. However, current physical behavior-based anomaly detection methods fail to differentiate between cyber-attacks and physical faults in building energy systems. Moreover, the challenge in collecting real-world threat data with ground truth has led researchers to rely on numerical models with user-defined assumptions, which may not accurately reflect real-world conditions due to the lack of in-situ experimental datasets. To address these challenges and gaps, this paper presents a flexible hardware-in-the-loop (HIL) testbed for generating cyber-attack and physical fault datasets and demonstrating threat detection algorithms in a real building automation system (BAS) environment. This testbed combines hardware (i.e., real BAS with local HVAC controllers and a physical network) with software (i.e., high-fidelity models to represent behaviors of building envelope and HVAC energy systems), enabling emulations of realistic threats. Five HIL experiments, including one baseline without any threats, two with physical faults, and two with cyber-attacks, were conducted to generate datasets containing detailed network traffic and system states. A joint classification framework, incorporating a network analyzer and a physical HVAC fault detector, was proposed to automatically detect cyber-physical abnormalities on BAS at both the network and the physical HVAC levels. The network analyzer comprises a conditional random fields (CRF) based command validator and a statistics-based detection strategy. The fault detector employs a weather and schedule-based pattern matching and feature-based principal component analysis (WPM-FPCA) method. Evaluation of the classification using four metrics from the multi-class confusion matrix revealed an average accuracy of 90.2%, recall of 89.7%, precision of 88.5% and F1-score of 89.2%. Finally, these results demonstrate that the proposed joint classification framework can effectively differentiate between specific types of cyber-attacks (e.g., device reinitialization attack, network Denial-of-Service attack) and physical faults (e.g., air handling unit operational fault, cooling coil valve stuck) in real time for improved building energy management.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2018

The geophysics, geology and mechanics of slow fault slip

We report that modern geodetic and seismologic observations describe the behavior of fault slip over a vast range of spatial and temporal scales. Slip at sub-seismogenic speeds is evident from top to bottom of lithospheric faults and plays an important role throughout the earthquake cycle. Where earthquakes and tremor accompany slow slip, they help illuminate the spatiotemporal evolution of fault slip. Geophysical subsurface imaging and geologic field studies provide information about suitable environments of slow slip. In particular, exhumed fault and shear zones from various depths reveal the importance of multiple deformation processes and fault-zone structures. Most geologic examples feature frictionally weak and velocity-strengthening materials, well-developed mineral fabrics, and abundant veining indicative of near-lithostatic fluid pressure. To produce transient slow slip events and tremor, in addition to the presence of high-pressure fluids a heterogeneous fault-zone structure, composition, and/or metamorphic assemblage may be needed. Laboratory and computational models suggest that velocity-weakening slip patches smaller than a critical dimension needed for earthquake nucleation will also fail in slow slip events. Changes in fluid pressure or slip rate can cause a fault to transition between stable and unstable fault slip behavior. Lastly, future interdisciplinary investigations of slow fault slip, directly integrating geophysical, geological and modeling investigations, will further improve our understanding of the dynamics of slow slip and aid in providing more accurate earthquake hazard characterizations.

58 GEOSCIENCES↗
DOE OSTI2020

Hydraulic characterization of a fault zone from fracture distribution

A quantitative assessment of how faults control the migration of geofluids is critical in many areas of geosciences. Here we integrated geological fieldwork, quantitative analysis of the fractures distribution and numerical modeling to build a geometrical representation of a fault zone and to characterize its hydraulic properties. Our target is a fault located in the Majella Mountain (Italy). We collected 21 scan lines across the fault profile in order to characterize its architecture. The numerical modeling of the fracture network of the damage zones and their hydraulic parameters was performed using both commercial (Move ® ) and open source software (dfnWorks and PFLOTRAN). Move ® was used to build a representative model of the fault zone using fracture spacing as a proxy, and to model the hydraulic parameters of the different fault domains. dfnWorks and PFLOTRAN were employed to infer the hydraulic parameters of the damage zones of the fault and then upscale these properties to an equivalent continuum domain, suitable for fluid flow simulations through the whole fault zone. Our findings show how even in a relatively small area it is possible to describe changes in terms of hydraulic properties of a fault zone and to build models capable to represent these variations.

58 GEOSCIENCES↗
DOE OSTI2021

Effect of Glacial/Interglacial Recharge Conditions on Flow of Meteoric Water Through Deep Orogenic Faults: Insights Into the Geothermal System at Grimsel Pass, Switzerland

Many meteoric-recharged, fault-hosted geothermal systems in amagmatic orogenic belts have been active through the Pleistocene glacial/interglacial climate fluctuations. The effects of such climate-induced recharge variations on fluid flow patterns and residence times of the thermal waters are complex and may influence how the geothermal and mineralization potential of the systems are evaluated. We report systematic thermal–hydraulic simulations designed to reveal the effects of recharge variations, using a model patterned on the orogenic geothermal system at Grimsel Pass in the Swiss Alps. Fault-bound circulation of meteoric water through the Grimsel fault is driven to depths of ~10 km by the high alpine topography. Simulations suggest that the current single pass flow is typical of interglacial periods, during which 1) meteoric recharge into the fault is high (above tens of centimeters per year), 2) conditions are at or somewhat below the critical Rayleigh number, and 3) the hydraulic connectivity along the fault plane is extensive (an extent of at least 10 km into increasingly higher terrain is required to explain the 10 km penetration depth). The subcritical condition constrains the bulk fault permeability to < 1e-14 m2. In contrast, the limited recharge during the numerous Pleistocene glaciation events likely induced a layered flow system with single pass flow confined to shallow depths while non-Rayleigh convection occurred in the deep fault. The same layering can be observed at low aspect ratios (length/depth) of the fault plane, when the available recharge area limits the flux through the fault.

58 GEOSCIENCES↗
DOE OSTI2023

Active Fault Current Limiting Control for Half-Bridge MMC in HVDC Systems

DC faults of MMC can result in a significantly large fault current due to the discharge of submodule capacitors. The fault current not only risks damaging the MMC but also demands a considerable breaking capacity from the dc circuit breaker (DCCBs). This paper introduces two novel active fault current limiting methods (AFCLs), namely virtual impedance-based and energy control-based AFCL. The first method utilizes circulating current feedforward, which introduces a virtual arm impedance to suppress the rate of rise of the fault current. Meanwhile, the second method relies on the control of the internally stored energy of the MMC to automatically minimize the number of submodules that discharge during a dc-side fault. Therefore, both the dc-side current and the MMC arm current can be effectively suppressed after the occurrence of the fault. The proposed methods do not require fault detection and their response is proportional to the rate of rise in the fault current. Simulation case studies are presented to demonstrate the proposed methods.

active current limiting control↗
DOE OSTI2016

Fault Modeling of Extreme Scale Applications Using Machine Learning

Faults are commonplace in large scale systems. These systems experience a variety of faults such as transient, permanent and intermittent. Multi-bit faults are typically not corrected by the hardware resulting in an error. Here, this paper attempts to answer an important question: Given a multi-bit fault in main memory, will it result in an application error — and hence a recovery algorithm should be invoked — or can it be safely ignored? We propose an application fault modeling methodology to answer this question. Given a fault signature (a set of attributes comprising of system and application state), we use machine learning to create a model which predicts whether a multibit permanent/transient main memory fault will likely result in error. We present the design elements such as the fault injection methodology for covering important data structures, the application and system attributes which should be used for learning the model, the supervised learning algorithms (and potentially ensembles), and important metrics. Lastly, we use three applications — NWChem, LULESH and SVM — as examples for demonstrating the effectiveness of the proposed fault modeling methodology.

97 MATHEMATICS AND COMPUTING↗