In-service Fault Research Articles

Data-driven approach for labelling process plant event data

An essential requirement in any data analysis is to have a response variable representing the aim of the analysis. Much academic work is based on laboratory or simulated data, where the experiment is controlled, and the ground truth clearly defined. This is seldom the reality for equipment performance in an industrial environment and it is common to find issues with the response variable in industry situations. We discuss this matter using a case study where the problem is to detect an asset event (failure) using data available but for which no ground truth is available from historical records. Our data frame contains measurements of 14 sensors recorded every minute from a process control system and 4 current motors on the asset of interest over a three year period. In this situation the ``how to'' label the event of interest is of fundamental importance. Different labelling strategies will generate different models with direct impact on the in-service fault detection efficacy of the resulting model. We discuss a data-driven approach to label a binary response variable (fault/anomaly detection) and compare it to a rule-based approach. Labelling of the time series was performed using dynamic time warping followed by agglomerative hierarchical clustering to group events with similar event dynamics. Both data sets have significant imbalance with 1,200,000 non-event data but only 150 events in the rule-based data set and 64 events in the data-driven data set. We study the performance of the models based on these two different labelling strategies, treating each data set independently. We describe decisions made in window-size selection, managing imbalance, hyper-parameter tuning, training and test selection, and use two models, logistic regression and random forest for event detection. We estimate useful models for both data sets. By useful, we understand that we could detect events for the first four months in the test set. However as the months progressed the performance of both models deteriorated, with an increasing number of false positives, reflecting possible changes in dynamics of the system. This work raises questions such as ``what are we detecting?'' and ``is there a right way to label?'' and presents a data driven approach to support labelling of historical events in process plant data for event detection in the absence of ground truth data.

Online Fault Detection and Isolation in Advanced Heavy Water Reactor Using Multiscale Principal Component Analysis

Signals from sensors placed at different locations are used for control and protection purposes in a nuclear reactor. It also requires in-service fault detection and isolation (FDI) for its safe operation. The sensor signals are generally a superimposition of low-frequency components representative of true values of the variables being monitored, occasional high-frequency periodical oscillations due to disturbances and faults, and sensor faults. Techniques like principal component analysis (PCA) can be used for FDI; however, it would be more meaningful if the FDI technique can also help for predictive maintenance of reactor internals through vibration spectra. To address these issues, a multiscale PCA, integrating wavelet transform with PCA and aiming to reduce the modeling cost by using only a fewer scales that contribute to the monitoring, has been proposed in this paper for online FDI of an advanced heavy water reactor (AHWR). A new mathematical formulation of the generalized likelihood ratio test for its use with wavelet approximation coefficients has also been proposed for better sensor-FDI outcomes. The proposed approach detects and isolates sensor faults and process faults using the signals from neutron detectors. Efficacy of the proposed technique is established on the simulated ex-core ion chamber data of AHWR considering different scenarios that involve localized frequency contents representative of process faults, slowly developing (incipient) sensor faults, and the simultaneous presence of two or more of these scenarios. Simulation results validate the effectiveness of the proposed scheme for online FDI in the reactor.

A Low-Frequency Tone Sweep Method for In-Service Fault Location in Subcarrier Multiplexed Optical Fiber Networks

We demonstrate an optical fiber fault location method based on the frequency response of the modulated fiber optical backscattered signal in a steady state low-frequency step regime. Careful calibration and measurement allows for the reconstruction of the fiber transfer function, which, associated to its mathematical model, is capable of extracting the fiber characteristics. The technique is capable of identifying non-reflective fault events in an optical fiber link and is perfectly compatible with previous methods that focus on the reflective events. The fact that the recuperation of the complex signal is performed in the frequency domain and not via a Fourier Transform enables the measurements to overcome the spatial resolution limitation of Fourier Transform incoherent-OFDR measurements even with frequency sweep ranges down to 100-100000 Hz. This result is backed up by a less than 10 meters difference in fault location when compared to standard OTDR measurements.

Distribution drop fiber in-service fault management in enhanced EPON system

Ethernet passive optical network (EPON) is considered one of the best candidates for next-generation optical access solutions. Practical and cost-effective survivability and maintenance mechanisms are therefore becoming the key issues for continued development of viable EPON solutions. In addition, 80% of faults occur within the first/last mile. Therefore, in this paper, we propose in-service integrated fault management mechanisms at the distribution drop fibers (DDFs) in EPON. The proposed mechanisms can automatically detect and identify any signal anomalies and DDF faults in the physical layer, thus decreasing the operating expense (OpEx) to be incurred in sending off technicians in the field. The mechanisms are two-fold: pre-fault and post-fault. In the pre-fault mechanism, we use optical receiver type/length/value (TLV) code as a sensor for triggering the embedded miniaturized optical-time-domain reflectometry (OTDR), placed at optical network units (ONUs). This mechanism observes and measures the abnormal condition and reports it to the optical line terminal (OLT), via restoration plan fiber link. In post-fault mechanism, we divide the ONUs into several restoration groups (RGs) to guarantee upstream transmission of the affected ONUs. Lastly, the fault dynamic bandwidth allocation for fault management, referred as FDBA, is presented. Simulation results show that the proposed mechanisms can withstand DDF faults without any significant impact on the overall quality of service (QoS) performance, in terms of mean packet delay, queue length, system throughput, and packet loss.

Survivable WDM-PON with self-protection and in-service fault localization capabilities

In this paper, we analyze the survivability of wavelength division multiplexed (WDM) passive optical network (PON). We propose and experimentally demonstrate a novel survivable WDM-PON architecture with self-protection and in-service fault localization capabilities. Two WDM-PONs are combined together and multiple sub-ring architecture is formed. Protection switching is achieved automatically at optical network units (ONUs) when failure occurs. Fiber loss, fiber fault localization and automatic protection switching in each ONU are real-time monitored in central office (CO) without interrupting customer service. No signaling for failure notice or protection switching is required for either CO or ONUs. Extra fibers required for protection are minimized compared with previous protection schemes. The self-protection and in-service fault localization capabilities of the proposed WDM-PON architecture are experimentally verified.

Fiber-fault monitoring technique for passive optical networks based on fiber Bragg gratings and semiconductor optical amplifier

A novel monitoring method for in-service fault indemnification in passive optical networks (PONs) is proposed and demonstrated experimentally. The proposed monitoring technique is based on fiber Bragg grating (FBG) sensors and fiber laser scheme, and a semiconductor optical amplifier (SOA) is used to serve as a gain medium. By detecting the number of the wavelength lasing, the fiber-fault can be monitored without affecting the in-service channels. Moreover, the strain behavior on the FBGs has also been discussed.

Optical fiber-fault surveillance for passive optical networks in S-band operation window

An S-band (1470 to 1520 nm) fiber laser scheme, which uses multiple fiber Bragg grating (FBG) elements as feedback elements on each passive branch, is proposed and described for in-service fault identification in passive optical networks (PONs). By tuning a wavelength selective filter located within the laser cavity over a gain bandwidth, the fiber-fault of each branch can be monitored without affecting the in-service channels. In our experiment, an S-band four-branch monitoring tree-structured PON system is demonstrated and investigated experimentally.

A novel in-service surveillance scheme for optically amplified transmission systems

We propose here a novel surveillance scheme for an optically amplified transmission system which allows simultaneous in-service fault identification for fiber and optical amplifiers. While requiring additional fiber Bragg gratings of designated wavelengths to be integrated with optical amplifiers, the monitoring light source is derived from the unused spectra of ASE and no dedicated light source is needed as in conventional OTDR technique. We have further demonstrated the scheme's feasibility with a 100-km, three-EDFA optical transmission system.

In-service passive surveillance system for optically-amplifiedbranched networks

The authors experimentally demonstrate a passive surveillance system for in-service fault identification in optically-amplified branched networks. The proposed scheme only requires fibre Bragg gratings to be placed on different fibre branches and needs no extra laser source as required in conventional and multi-wavelength OTDR. The reflected monitoring channels are demultiplexed by an array-waveguide grating and detected by a power sensor array. The link status of all fibre branches can be monitored continuously and simultaneously through the Internet.

A practical passive surveillance scheme for optically amplified passive branched optical networks

We propose and experimentally demonstrate a novel and practical surveillance scheme for in-service fault identification in passive optically amplified branched networks. This scheme does not require any wavelength-tunable light source as used in multiwavelength optical time-domain reflectometer (OTDR). Fiber-Bragg gratings are placed at some strategic positions on different fiber branches to slice and reflect the optical amplifier's residual amplified spontaneous emission (ASE) power at some wavelengths other than the signal wavelengths. The conditions of the fiber link and optical amplifier at each branch can be monitored by constantly checking the reflected power level of the corresponding wavelength without suspending the in-service channels.

Proposal of decision support system for cooperation in hierarchical structures

In most previous research on group decision making systems, all participants are assumed to be working on the same decision. But, in electric power systems and the like, in-service fault handling must deal with a variety of externally caused problems in real time. The system described here assumes a hierarchical organization in which the upper levels issue broad strategic directives and the lower levels take specific actions based on them. Even in the making of seemingly diverse decisions, this decision process is actually cooperative. It requires a correct comprehension of system states and of the intentions of others (goals and strategies). A representation that superimposed goals, strategies, and states of the system is capable of revealing inconsistencies between intentions and circumstances. The proposed system has undergone expert evaluation in applications to two malfunctions in an electric power system. The representation method promotes a focus on errors in strategy and in the transmission of strategy from upper to lower levels and facilitates the revision of strategy in case of error. © 1997 Scripta Technica, Inc. Electron Comm Japan Pt 3, 80(2): 78–91, 1997

New method of in-service fault localisation in passive optical subscriber loops

A new method of detecting fibre breaks in a passive optical subscriber network without interruption to nonaffected customers is demonstrated. For these live measurements, an optical frequency domain reflectometry technique (OFDR) is used. No separate wavelength or additional components in the network are required. An integration facility on the AM/VSB TV-transmission system, without affecting TV-picture quality, is realised.

Implanted Component Faults and Their Effects on Gas Turbine Engine Performance

Under the sponsorship of the Canadian Department of National Defence, the Engine Laboratory of the National Research Council of Canada (NRCC) has established a program for the evaluation of component deterioration on gas turbine engine performance. The effect is aimed at investigating the effects of typical in-service faults on the performance characteristics of each individual engine component. The objective of the program is the development of a generalized fault library, which will be used with fault identification techniques in the field, to reduce unscheduled maintenance. To evaluate the effects of implanted faults on the performance of a single spool engine, such as an Allison T56 turboprop engine, a series of faulted parts were installed. For this paper the following faults were analyzed: (a) first-stage turbine nozzle erosion damage; (b) first-stage turbine rotor blade untwist; (c) compressor seal wear; (d) first and second-stage compressor blade tip clearance increase. This paper describes the project objectives, the experimental installation, and the results of the fault implantation on engine performance. Discussed are performance variations on both engine and component characteristics. As the performance changes were significant, a rigorous measurement uncertainty analysis is included.

Arc motion and burn through in GIS

It is pointed out that two different mechanisms are effective for the burn-through process of GIS (gas insulated substation) encapsulation by an internal arc during an in-service fault: the material erosion by melting and evaporation and the weakening of the material by heating. Depending on the arc motion one of these mechanisms may be dominant. Hitherto inexplicable motions of the arc are explained; especially the arc rotation at a spacer and the axial motion in case of symmetrical current feeding. These considerations are confirmed by experimental results and experience from industry, and they result in some recommendations for the design and the protection concept. >

First field demonstration of in-service fault location/supervisory using optical time domain reflectometry

In-service fault location using optical time domain reflectometry is demonstrated for the first time over an installed submarine, fibre-optic cable. Results are presented showing simultaneous fault location and 565 Mbit/s data transmission on one fibre of the UK-Channel Islands no. 7 fibre-optic cable. In addition, laboratory experiments indicate that the technique could prove to be invaluable in supervising wavelength-routed networks incorporating both passive and active optical components.

Techniques and Instrumentation for Measurement of Transients in Gas-Insulated Switchgear

The measurement of transient phenomena in gasinsulated switchgear (GIS) has become a problem of interest to both utilities and manufacturers. Such transients have been responsible for a significant number of in-service faults and will probably be the critical parameter for dimensioning of GIS for system voltages over 550 kV. This paper describes the design, construction, and use of planar capacitive dividers and measuring systems which achieve subnanosecond rise times, response to below 1 mHz, and constant division ratio as a function of frequency.

Fault Location in SF6 Insulated Substations Using Thermal Techniques

This paper presents the results of a study of non-invasive fault location techniques in gas insulated substations. The applications of infrared thermal imaging system and temperature sensitive paint to locate in-service faults with power follow-through were demonstrated. The infrared fault location system can be applied to existing substation without retrofitting. Detection sensitivities and system parameters are presented.

Estimation of Burn-Through Probability in SF6 Insulated Substations

Enclosure burn-through in SF6 insulated substations is a rarely occurring event; however, in view of the possibly damaging effects to equipment and personnel, its probability of occurrence warrants a closer examination. This paper presents a model for estimating the likelihood of enclosure burn-through or pressure relief disc bursting during an in-service fault in SF6 substations. Results of a simulation show a 25% burn-through probability per fault in substations with the most probable fault currents in the 50 kA range. The probability of personnel being exposed to the hazardous arcing by-products is discussed.

Failure experience as a tool for evaluation of nuclear structures

High reliability in nuclear structures is attainable only because previous failure experience with comparable installations can be factored into structural evaluations in such a way that unsatisfactory experience is forestalled. The use of corrective techniques to eliminate in-service faults in consumer products as a means of improving the characteristics of future products is a well-known quality control principle. This principle is only applicable to nuclear structures if the nature and causes of structural failure are fully understood. The development of some well-established structural practices that evolved through the correlation of failure experience with structural evaluation methods is discussed in this paper. Attention is then directed to the use of analogous experience in nuclear applications as an approach to the attainment of suitable structural failure characteristics. A recognized shortcoming in the use of failure experience as a basis for judging structural adequacy is the paucity of statistical data to support the evaluation process. The use of calculational and experimental methods that simulate failure characteristics as a substitute for real experience is a valuable adjunct to such statistical data. The correlation between actual failure experience and simulation techniques is set forth as a crucial feature of nuclear structural evaluations with careful interpretation of actual failure causes and effects as the prime control on structural evaluation methods.

Cable in Trouble?

This paper presents the problems encountered in the underground medium-voltage distribution system at a large manufacturing facility of Western Electric Company, Inc. Three inservice faults and subsequent test failures were experienced within a 5-month period in 1963. A dc overpotential test program was initiated, as a result of which, the decision was made to replace the entire underground system.