On-line Predictive Maintenance Research Articles

Designing a Resilient Building Maintenance Program: Integrating Preventive and Predictive Maintenance at the U.S. Navy

ABSTRACT Building equipment failure can have drastic effects on a company's operations and budget. This paper presents two types of maintenance approaches that if done effectively, can prevent or significantly reduce the failure of building equipment assets. The first is traditional time-based preventive maintenance (PM), which conducts prefailure inspections and tasks in a cyclic time-based approach. The second, is predictive maintenance (PdM), which conducts maintenance functions based on the condition of the equipment found through continuous or cyclic measurements and analysis during machine operation. The purpose of investigating these maintenance approaches is to determine whether we can improve on the U.S. Navy's (Navy) existing facility maintenance program, helping to reduce overall costs while improving sustainability, equipment resiliency, and efficiency. By presenting each maintenance program and leveraging today's technologies we show that these advancements in technology can directly improve the Navy's operational mission and warfighter readiness. Research was conducted through the following methods: interviews, books, third party reports, journal articles, industry websites and articles that focus on equipment maintenance. The Navy, the National Aeronautics and Space Administration (NASA), and the University of Washington provided case studies, existing facility/utility maintenance data, and budget information used in this research. The results show that PdM approaches using advanced analytics are more effective in diagnosing equipment, prescribing equipment problems, and predicting equipment failure. It will also show that when a PdM model is used, building tenants have less operational impacts as equipment operates longer with less downtime between maintenance events. By changing to a PdM program, facility managers and owners can improve asset efficiency and resilience, directly improving environmental sustainability and lowering overall longterm costs. It highlights the significant capital costs of a fully online PdM program and the benefits of using a hybrid model of PM and PdM. This research concludes with an overview of how building maintenance is currently being conducted on Navy bases by Naval Facilities Engineering Systems Command (NAVFAC), and how they are transitioning to a more sustainable maintenance program leveraging existing advanced metering infrastructure (AMI), building control systems (BCS), and utility control systems (UCS) with Smart Grid (SG), OSI Pi, and advanced analytics. In addition, major gaps in this transition are identified, and solutions are proposed to optimize these building system investments.

Development of an energy harvesting system based on a thermoelectric generator for use in online predictive maintenance systems of industrial electric motors

In the era of the Fourth Industrial Revolution, improvements in maintenance routines for industrial electric motors are continuously being made, allowing for the development of Predictive Maintenance (PdM) Online. PdM can be executed by employing small embedded systems that are affixed to electrical machines. Although a small-capacity battery can effectively power an ultra-low consumption sensor, the extensive use of batteries can contribute to increased environmental pollution. An effective approach to obviate the necessity for batteries is to employ energy-harvesting techniques from electric motors. In this regard, this paper aims to contribute to advancements in research concerning the harvesting of thermoelectric energy from industrial electric motors to create a self-sustaining Internet of Things (IoT) device. To achieve energy autonomy, the construction of a mechanical coupling for the attachment of Thermoelectric Generators (TEGs) was proposed, along with the maximization of small TEG production and the optimization of the electronic circuit for energy management. The mechanical prototype for affixing TEGs to low-power electric motors enabled a maximization of the generated energy by the TEGs, and the electronic circuit optimization process resulted in the generation of 370.10 μW, exceeding the energy requirements for making the IoT device self-sustaining by 54.21%.

On-line quality control and tool wear evaluation in trimming process by data analytics techniques

Sheet metal is an extremely common form of use in broad engineering sectors and represents the main consumption of steel and aluminum in Europe. All production processes where sheet metal is used require cutting operations. Therefore, it is an extremely relevant procedure, common and of great economic importance in material-intensive sectors such as automotive, construction and engineering. In a competitive environment where the demand for quality increases continuously and using harder to process high-performance materials, online monitoring and quality control become an attractive prospective. In this regard, application of data analytics techniques offers a promising toolset for fast and efficient analysis. This work presents a pilot demonstration of an instrumented sheet metal cutting line that integrates online quality control and predictive maintenance concepts, namely detection of wear in tools as well as problems arising from the raw material and process. Automated sequential cutting was performed using a straight trimming die mounted on a hydraulic press to cut High-Mn TWIP steel strip. Three piezoelectric force sensors were located in the trimming tool, collecting cycle data. Tests performed included studying the effect of blunting on the M2 (1.3343) tool, as well as the influence of correct lubrication. The resulting data were analyzed through the extraction of several parameters from each cycle to correlate die wear-induced force response evolutions with the shear-induced deformation in High-Mn TWIP steel sheets. Extracted parameters were the maximum and minimum force of each cycle and the impulse (integrated force over time) was computed and segmented by plastic deformation and crack contributions. Our analysis show that advanced algorithms can be used to analyze sensor data on an automated installation, revealing OK/NOK conditions and anomalous events.

Augmented model-based framework for battery remaining useful life prediction

Traditional, model-based approaches for predicting the remaining useful life (RUL) of a rechargeable battery cell simply update and extrapolate a mathematical model which describes the evolution of the cell’s capacity fade trend. These approaches are straightforward but tend to break down when the capacity fade trend changes over the cell’s lifetime. To retain the desirable properties of model-based prediction approaches (uncertainty quantification, long-term accuracy, limited physical meaning) and improve their overall accuracy in RUL prediction, we augment empirical model-based prediction with data-driven error correction. Our approach decomposes the task of RUL prediction into two steps: 1) Offline training of data-driven models for RUL error correction and 2) Online data-driven correction of model-based RUL prediction. The approach is evaluated on five datasets consisting of 237 cells: 1) three open-source datasets, 2) one proprietary dataset, and 3) a simulated out-of-distribution dataset. Results show that data-driven error correction effectively reduces root-mean-square-error by 40% and mean uncertainty calibration error by 34% compared to a model-based approach alone. The proposed approach is also shown to be more conservative in its uncertainty estimates than a purely data-driven RUL prediction approach. Special attention is given to ensure the initial model-based uncertainty estimates are propagated through the data-driven error correction model and considered in the final RUL prediction. The enhanced uncertainty quantification of our approach makes it suitable for deployment in an online predictive maintenance scheduling framework.

Principal component analysis technique for early fault detection

Online condition monitoring and predictive maintenance are crucial for the safe operation of equipments. This paper highlights an unsupervised statistical algorithm based on principal component analysis (PCA) for the predictive maintenance of industrial induced draft (ID) fan. The high vibration issues in ID fans cause the failure of the impellers and, sometimes, the complete breakdown of the fan-motor system. The condition monitoring system of the equipment should be reliable and avoid such a sudden breakdown or faults in the equipment. The proposed technique predicts the fault of the ID fan-motor system, being applicable for other rotating industrial equipment, and also for which the failure data, or historical data, is not available. The major problem in the industry is the monitoring of each and every machinery individually. To avoid this problem, three identical ID fans are monitored together using the proposed technique. This helps in the prediction of the faulty part and also the time left for the complete breakdown of the fan-motor system. This helps in forecasting the maintenance schedule for the equipment before breakdown. From the results, it is observed that the PCA-based technique is a good fit for early fault detection and getting alarmed under fault condition as compared with the conventional methods, including signal trend and fast Fourier transform (FFT) analysis.

Predictive maintenance of systems subject to hard failure based on proportional hazards model

The remaining useful lifetime (RUL) estimated from the in-situ degradation data has shown to be useful for online predictive maintenance. In the literature, the RUL is often estimated by assuming a soft-failure threshold for the degradation data. In practice, however, systems may not be subject to the degradation-induced soft failures. Instead, the systems are deemed to be fail when they cannot perform the intended function, and such failures are known as hard failures. Because there are no fixed thresholds for hard failures, the corresponding RUL estimation is not an easy task, which causes difficulties in finding the optimal maintenance schedule. In this study, a Weibull proportional hazards model is proposed to jointly model the degradation data and the failure time data. The degradation data are treated as the time-varying covariates so that the degradation does not directly lead to system failures, but increases the hazard rate of hard failures. A random-effects Wiener process is proposed to model the degradation data by considering the system heterogeneities. Based on the developed proportional hazards model, closed-form distribution of the RUL is derived upon each inspection and the optimal maintenance schedule is then obtained by minimizing the system maintenance cost. The proposed maintenance strategy is successfully applied to predictive maintenance of lead-acid batteries.

Towards intelligent and sustainable production systems with a zero-defect manufacturing approach in an Industry4.0 context

The paper addresses intelligent and sustainable production achieved through combination and integration of online predictive maintenance, monitoring of process parameters and continuous quality control of both input materials and output from the process. This enables production systems, within both manufacturing and process industries, to move towards zero-defect manufacturing. Such a zero-defect manufacturing approach allows for earlier identification of problems or issues, which will or already negatively affect the output. The paper outlines the first part of the second cycle of an action research effort at Gestamp HardTech AB in Sweden, whose objective is to keep its position as a world-leading provider of press-hardened vehicle parts. In order to fully implement the zero-defect manufacturing approach, 4-6 action research cycles are expected to be needed in order to iteratively refine the approach. During the first cycle, various methods and solutions for some of the individual issues/problems have been conceptualized, realized and initially tested. The selected design criteria for the action research efforts were: simplicity, low cost, robustness, high-quality output and future-proofing. The result from the research in the second cycle so far is an action plan for the technical change and a set of challenges/problems which need additional investigation.

Towards Intelligent and Sustainable Production: Combining and Integrating Online Predictive Maintenance and Continuous Quality Control

The paper addresses intelligent and sustainable production in the sense of combining and integrating online predictive maintenance and continuous quality control. The rationale for combining and integrating them is that continuous quality control can provide input to the online predictive maintenance in cases where no signs of maintenance issues have been indicated and inadequate output is produced (and the process parameters cannot be adjusted in order to meet the output specifications). The paper outlines the first part of an action research effort at Gestamp HardTech AB in Sweden, whose objective is to keep its position as a world-leading provider of press-hardened vehicle parts. The first initial design criteria concerned with simplicity and low-cost was changed after learning more details and what can be accomplished, and the updated first design criteria focused instead on robustness, high-quality output and future-proofing. The intermediary result is an action plan for the technical change, whereas the organizational change will be addressed later during the action research effort.

Application of IoT concept on predictive maintenance of industrial equipment

The Internet of Things (IoT) concept describes an intelligent connectivity of smart devices using the internet network. Nowadays, companies try different approaches for predictive maintenance as a solution to reduce costs and the frequency of maintenance activities. The IoT platforms provide a good support for predictive maintenance as it can integrate information from different machines and manufacturing systems. The main drawback in integrating production system with IoT dedicated platforms is the communication framework, knowing that the main industrial communication protocols are incompatible with modern communication protocols implemented on IoT platforms. In this context, the present paper proposes a new and simple method for on-line monitoring and predictive maintenance of industrial equipment. This method has two features of connected manufacturing. One of these is process monitoring for constant quality assurance, the other one is condition monitoring in order to prevent unplanned downtimes. A case study is presented to demonstrate the feasibility of the proposed method.

Implementation of a novel double-side technique for partial discharge detection and location in covered conductor overhead distribution networks

Partial discharge (PD) detection has proven to be one of the most acceptable techniques for on-line condition monitoring and predictive maintenance of power apparatus. A powerful tool for detecting PD in covered-conductor (CC) lines is urgently needed to improve the asset management of CC overhead distribution lines. In this paper, an appropriate, portable and simple system designed to detect PD activity in CC lines and ultimately pinpoint the PD source is developed and tested. The system is based on a novel double-side synchronised PD measurement technique driven by pulse injection. Emphasis is placed on the proposed PD-location mechanism and hardware structure, with descriptions of the pulse-injection process, detection device, synchronisation principle and PD-location algorithm. The system is simulated using ATP-EMTP, and the simulated results are found to be consistent with the actual simulation layout. For further validation, the capability of the system is tested in a high-voltage laboratory experiment using a 10-kV CC line with cross-linked polyethylene insulation.

Wavelet-based de-noising of on-line PD signals captured by Pearson coil in covered-conductor overhead distribution networks

Partial discharge (PD) detection has been regarded as one of the most effective on-line predictive maintenance test and diagnostic tool for the condition monitoring of high voltage (HV) equipments. The relatively new and challenging application is conducting on-line high frequency PD measurements for the monitoring of falling trees on the covered-conductor (CC) overhead distribution lines. Generally, the high frequency PD measurements taken in high voltage (HV) laboratory Faraday cage are not prone to electromagnetic disturbances (EMDs). However, various external interferences may corrupt on-line and on-site measurements. Getting very low magnitude PD signals from measurements taken in corrupt environment is one of the big challenges for continuous monitoring. In this paper, wavelet transform (WT) is suggested as a powerful technique to remove noise and disturbances from PD signals captured in CC overhead distribution networks, which are completely absorbed by electromagnetic interferences (EMIs). The advantages and difficulties faced while taking on-line and on-site PD measurements have been discussed. Pearson coil is used as a sensor to detect PDs for this specific application. In the next stage of future developments, the proposed methodology would be implemented in a real-life environment using many Pearson coils to get reliable on-line and on-site PD measurements to detect and localize falling tress on CC overhead distribution networks. The condition monitoring and insulation diagnosis of the other valuable distribution network components can also be carried out using this technique to cope with the requirements of future smart grids.

Calibration of on-line partial discharge measuring system using Rogowski coil in covered-conductor overhead distribution networks

Partial discharge (PD) detection has been regarded as one of the most effective on-line predictive maintenance test and diagnostic tool for the condition monitoring of high voltage (HV) equipment. In this study, a methodology for detecting on-line PDs produced because of falling trees on the covered-conductor (CC) overhead distribution lines is introduced and calibration of PD measuring system is carried-out. The Rogowski coil is used as a PD sensor which is non-intrusive and superior to the conventional PD detectors. The experimental set-up was arranged in the HV laboratory for real-time analysis and a pulse calibrator was used to calibrate the PD measuring system. Few real-life PD measurements have been taken and it is revealed that PD magnitudes and signals bandwidth may vary under various circumstances. The calibrated on-line PD measuring system can be used to detect and measure the amount of PDs produced because of falling trees on CC lines, thus improving the reliability and safety of the distribution networks as well as reducing visual inspection work after storms.

Wireless sensors for predictive maintenance of rotating equipment in research reactors

In 2008–2009, the High Flux Isotope Reactor (HFIR) at the Oak Ridge National Laboratory (ORNL) tested the potential of predictive or condition-based maintenance techniques to reduce maintenance costs, minimize the risk of catastrophic failures, and maximize system availability by attaching wireless-based sensors to selected rotating equipment at HFIR. Rotating equipment is an ideal “test case” for the viability of integrated, online predictive maintenance strategies because motors, bearings, and shafts are ubiquitous in nuclear power plants and because the maintenance methods typically performed on rotating equipment today (such as portable or handheld vibration data collection equipment) are highly labor-intensive. The HFIR project achieved all five of its objectives: (1) to identify rotating machinery of the types used in research reactors and determine their operational characteristics, degradation mechanisms, and failure modes, (2) to establish a predictive maintenance program for rotating equipment in research reactors, (3) to identify wireless sensors that are suitable for predictive maintenance of rotating machinery and test them in a laboratory setting, (4) to establish the requirements and procedures to be followed when implementing wireless sensors for predictive maintenance in research reactors, and (5) to develop a conceptual design for a predictive maintenance system for research reactors based on wireless sensors. The project demonstrated that wireless sensors offer an effective method for monitoring key process conditions continuously and remotely, thereby enhancing the safety, reliability, and efficiency of the aging research reactor fleet.

Partial discharge theory and technologies related to medium-voltage electrical equipment

Partial discharge (PD) monitoring is an effective online predictive maintenance test for medium-voltage (MV) motors, MV generators, and MV switchgear at 4160 V and above, as well as other electrical distribution equipment. The benefits of online testing allow for equipment analysis and diagnostics during normal production. Corrective actions can be planned and implemented, resulting in reduced unscheduled downtime. An understanding of the theory related to PD, and the relationship to early detection of insulation deterioration is required to properly evaluate this predictive maintenance tool. This paper presents a theory to promote the understanding of PD technology, as well as various implementation and measurement techniques that have evolved in the industry. Data interpretation, corrective actions, and application to electrical distribution equipment is also reviewed.

SYSTEM MODELING AND SIMULATION FOR PREDICTIVE MAINTENANCE

Condition monitoring, the process of data collection for the evaluation of machinery performance and reliability, is an essential part of today's on-line predictive maintenance for complex manufacturing systems. Together with the increasing complexity of machinery systems, the number of assets that need to be diagnosed is also rapidly increasing. This is why the management of condition monitoring became a complex cooperation and resource allocation problem. This paper describes an implementation of computer simulation as a modeling and decision support tool for the management of a condition-monitoring service group. The implementation was developed using the SLAMSYSTEM modeling environment. It provides an easily adaptable management-style program that may be applied to a company with a monitoring system that varies in versatility, expertise, and availability.

Performance analysis of rotating machinery using enhanced cerebellar model articulation controller (E-CMAC) neural networks

For today's sophisticated machinery systems, on-line predictive maintenance has become a most reliable and cost effective method for machinery maintenance. An effective fault monitoring and diagnosis tool is able to recognize the characteristics, conditions, and developing trends of an operating machinery system, and is able to estimate fault severity of the system quantitatively. In order to meet the requirements, a new approach combining cerebellar model articulation controller (CMAC) neural networks and advanced vibration monitoring methods has been presented. A test rig consisting of two rotating hubs driven by a d.c. motor was used to produce different machine imbalance levels. A two-stage experiment is proposed. First, a CMAC network is trained with pre-defined imbalance conditions and a test data set is given to check the trained CMAC for imbalance severity prediction. Second, a CMAC network is trained by using a set of different imbalance levels and is then given an untrained condition to test the CMACs ability of imbalance severity interpolation. In addition, several potential capabilities of CMAC networks will be discussed and shown. The properties of the CMAC are notable. If implemented properly, CMAC can be used as an effective machinery condition monitor and fault severity estimation tool.