Increase Machine Availability Research Articles

Concept of operation of the Agile Robot Transporter for ITER in-vessel maintenance

ITER machine availability is crucial for fusion reactors in order to be the main source of electricity, which is affected by many aspects. This paper focuses on maintenance aspects of the machine. As the plasma facing components inside the fusion reactors become highly activated caused by the neutrons generated by fusion reactions, maintenance of the fusion reactors needs to be done by remote means. In order to increase machine availability, it is required to have efficient maintenance scenario in terms of time and effort, which can be achieved through effective in-vessel remote handling (RH) operations, simplified maintenance of the RH system itself, well-coordinated logistics, and trained personnel, etc. All these aspects need to be investigated during the design phases, which requires a well-defined concept of operation of the RH systems. This paper describes the concept of operation of the Agile Robot Transporter (ART), an ITER in-vessel RH system, that is under development to carry out neutron and optical diagnostic calibrations, and removal/replacement of the blanket first wall samples. Based on the functional analysis of the required in-vessel tasks, system requirements are defined. Among them, key requirements to improve RH operation efficiency are identified and the ART is designed accordingly. Efficient deployment and removal operations of the ART are crucial so that it requires less vacuum vessel (VV) port opening and less cask transportation, which reduces needs for hands-on operations resulting in less human radiation exposure. Various and flexible tool exchange and storage concepts are proposed that allow parallel operations between the in-vessel and ex-vessel RH systems. Modular and re-configurable design is proposed for easy RH system maintenance, and effective rescue operations to rescue the RH system when it fails.

A memetic based algorithm for simultaneous preventive maintenance scheduling and spare-parts inventory management for manufacturing systems

To increase machine availability, it may be necessary to halt them to replace components before failures occur. Therefore, optimizing spare parts inventory management is a key factor in improving maintenance activities. Spare part costs are a significant contributor to total maintenance costs, including holding and ordering costs, and are often dependent on the maintenance schedule. Hence, it is important to optimize preventive maintenance scheduling and spare parts inventory management together. The first attempt to introduce a new Mixed-Integer Linear Programming (MILP) model that integrates both decisions was by Mjirda et al. (2016). Our paper is the first to extend their work in order to come up with an efficient aided decision making framework based on a mat-heuristic approach that delivers a good trade-off between computational time and solution quality. Indeed, this paper introduces a novel Hybrid Memetic Algorithm (HMA) that synergistically combines Memetic Algorithm (MA) with MILP to solve this NP-hard problem. The paper aims to provide a unified and efficient mat-heuristic based aided decision making framework that optimizes the sum of maintenance, operating, holding, and ordering costs by synchronizing preventive maintenance schedules and spare parts inventory for a set of independent machines. Its efficiency is illustrated through an extensive computational study. The HMA outperforms standalone MILP solvers, solving 36% of instances to optimality within a computational time of less than 141 s, and achieving a maximum gap of 5% for 46% of instances within less than 200 s. Our findings highlight the computational efficiency and solution quality offered by the HMA, thereby advancing the state-of-the-art in integrated spare parts inventory management and preventive maintenance scheduling.

Few-shot remaining useful life prediction based on meta-learning with deep sparse kernel network

Predicting remaining useful life (RUL) of machinery is of vital importance to prognostics and health management. Reliable and accurate RUL prediction not only can reduce maintenance costs and increase machine availability but also even prevent catastrophic consequences. In reality, RUL predictions usually require numerous certain kinds of machine degradation data. However, complex operating conditions and safety issues may often result in fragmented data records generated, with very few complete samples being usable. To overcome the challenge of RUL prediction with limited data, this paper proposes a novel MetaDESK model that is based on meta-learning with deep sparse kernel network. The general idea is to train a sparse kernel with a variational posterior in a data-driven fashion, and then transfer it to a new few-shot RUL task via meta-knowledge. Specifically, we first incorporate a Gaussian Process into the model-agnostic meta-learning (MAML) framework and use variational inference to estimate latent variables as kernel features, which allows us to sample from a non-Gaussian distribution of the posterior. Then, the KL-divergence of sparse approximation is added to the kernel features as a regularization term through inference to reduce the overfitting problem. Also, to exploit the dependencies of the tasks we integrate both their shared knowledge and task-specific information into a contextual reasoning process, which is implemented by a bidirectional long short-term memory network. To evaluate our proposed model, we conduct extensive experiments using publicly available degradation data, and the results verify the model's effectiveness.

Determination of Maintenance Intervals Using Multi-Criteria Approach to Increase Machine Availability at a Cement Company

A downtime will decrease machine availability that will affect reliability and maintainability. If interval time between two consecutive machine failures is approach to zero, it shows that the machine reliability is low, and so the availability. In the other hand, if interval time between is high, the machine availability and reliability are high. However, the procuring cost of the machine components is expensive. The problem is what is the maintenance interval time in order to optimize machine availability and reliability. In this study, the downtime is minimized by determining maintenance intervals using a multicriteria, i.e. reliability, availability, and maintainability (RAM). The model used in this research is the Age Replacement Model. A case study of a cement company is taken to show the determination of the maintenance interval. The observation shows that the Kiln Division has the longest total downtime of 1,846 hours/year although the Kiln Division has implemented overhaul during 15 days and maintenance one times. The total downtime of the Kiln Division was caused by failure to the Grate Cooler machine by 56.582%. A maintenance interval is proposed, and the result shows that the proposed maintenance intervals reduce total downtime by 86.6% and improve machine availability by 1,8%.

Health indicator for machine condition monitoring built in the latent space of a deep autoencoder

The construction of effective health indicators plays a key role in the engineering systems field: they reflect the degradation degree of the system under study, thus providing vital information for critical tasks ranging from anomaly detection to remaining useful life estimation, with benefits such as reduced maintenance costs, improved productivity or increased machine availability. The reconstruction error of deep autoencoders has been widely used in the literature for this purpose, but this approach does not fully exploit the hierarchical nature of deep models. Instead, we propose to take advantage of the disentangled representations of data that are available in the latent space of autoencoders, by using the latent reconstruction error as machine health indicator. We have tested our proposal on three different datasets, considering two types of autoencoders (deep autoencoder and variational autoencoder), and comparing its performance with that of state-of-the-art approaches in terms of well-known quality metrics. The results of the research demonstrate the capability of our health indicator to outperform conventional approaches, in the three datasets, and regardless of the type of autoencoder used to generate the residuals. In addition, we provide some intuition on the suitability of latent spaces for the monitoring of machinery condition.

Efficacy Study of Fault Trending Algorithm to Prevent Fault Occurrence on Automatic Trampoline Webbing Machine

Nowadays, fault diagnostics is widely applied under Industry 4.0 to reduce machine maintenance costs, improve productivity, and increase machine availability. However, fault diagnostics are mostly post-mortem. When the fault is identified, it is already too late because damages have been done to the product and machine. This paper compares the efficacy of several signal data processing techniques for detecting faults that are about to occur. Our aim is to find an efficient way to predict the fault before it occurs. A continuous wavelet transform synchrosqueezed scalogram was found to be most suitable for this purpose, but it is difficult to apply. A novel procedure is proposed to count the number of pulses in the synchrosqueezed scalogram. A new method for detecting the trend from the pulse counts is then developed to predict the fault before it happens. The procedure and method are illustrated with experimental data collected while running an automated double-thread trampoline webbing machine.

Machine Learning-Based Fault Diagnosis of Self-Aligning Bearings for Rotating Machinery Using Infrared Thermography

Bearings are considered as indispensable and critical components of mechanical equipment, which support the basic forces and dynamic loads. Across different condition monitoring (CM) techniques, infrared thermography (IRT) has gained the limelight due to its noncontact nature, high accuracy, and reliability. This article presents the use of IRT for the bearing fault diagnosis. A two-dimensional discrete wavelet transform (2D-DWT) has been applied for the decomposition of the thermal image. Principal component analysis (PCA) has been used for the reduction of dimensionality of extracted features, and thereafter the most relevant features are accomplished. Furthermore, support vector machine (SVM), linear discriminant analysis (LDA), and k-nearest neighbor (KNN) as the classifiers were considered for classification of faults and performance assessment. The results reveal that the SVM outperformed LDA as well as KNN. Noncontact condition monitoring shows a great potential to be implemented in determining the health of machine. The utilization of noncontact thermal imaging-based instruments has enormous potential in anticipating the maintenance and increased machine availability.

Artificial Wear for the Assessment of Monitoring Performance

Various tool condition monitoring systems exist, that can increase machine availability and process reliability. Assessing and comparing their performance, however, requires high expenditure due to real process failures being scarce, too different or costly to reproduce. Hence, this paper investigates the reproducible simulation of flank wear. It introduces and validates a geometry for indexable inserts that results in process changes similar to those caused by natural flank wear. The validation considers turning processes with different feeds, depth of cut and cutting speeds in steel. Results demonstrate that the proposed geometry for indexable inserts affects process forces similar to natural flank wear.

Lean Manufacturing Production Management Model using the Johnson Method Approach to Reduce Delivery Delays for Printing Production Lines in the Digital Graphic Design Industry

Several factors compel graphic design companies to improve efficiency and competitiveness in their production lines. However, these companies are not prepared to take on this challenge, as they report delays in 20% of their deliveries, caused by high setup times, low machine availability, and poor work scheduling. In this context, this study proposes a new production management model fed by the interaction of lean manufacturing tools and the Johnson scheduling method. This model has been validated by direct application at the SISSA. The results obtained were the reduction of the setup time to 15 minutes, increased machine availability up to 24%, and an efficient scheduling of its tasks. All of these reduced the percentage of delivery delays from 20% to 6%.

Maintenance Management Model under the TPM approach to Reduce Machine Breakdowns in Peruvian Giant Squid Processing SMEs

In recent years, the giant squid processing industry in Peru exhibited a 59% increase in exports with respect to 2018. According to estimates, this industry generates approximately 30,900 jobs per year. However, some SMEs experience low productivity, such as the PECEPE company, due to plant downtime. This represents 26% of the available time, which translates into the loss of 1760 tons every year. The most constraining external factor the sector faces is uncertainty in resource availability caused by changing weather conditions and informal fishing activities. Although there is a large number of research studies on the fishing industry and resource extraction, literature on processing plant operations is scarce. Within this context, this study seeks to promote a high impact sector in Peru, as well as fostering processing plant competitiveness and productivity. Hence, to address these issues, the authors propose a maintenance management model under the TPM approach. As part of the results from model implementation, a 39% decrease was reported in plant downtime, while maintenance costs also decreased by 16%, which, in turn, increased machine availability and production by 784 tons per year.

Fourier Harmonic Regression Method for Bearing Condition Monitoring using Vibration Measurements

Recently, condition monitoring and fault diagnosis of rolling element bearings in rotary machinery is an important task in industries for system preventive maintenance, increase reliability, improved productivity and increase machine availability. Existing methods for condition monitoring of bearings require extensive experimental arrangements using costly sensors and complex computational techniques in frequency domain for fault diagnosis. This work presents a simple time domain based approach using Fourier harmonic regression analysis method for condition monitoring of rolling element bearings. In the present work, bearing vibrations are acquired in normal and faulty conditions. A Fourier harmonic regression analysis method is applied to determine the confidence intervals for the bearing condition monitoring. The magnitude of harmonics and standard deviation of residuals are analyzed for identifying effect of faults like outer race fault, inner race fault and bearing ball fault. Experimental results show the effectiveness in detection of bearing conditions using the calculations in time domain.

Small synchronous machine protection during voltage sags caused by MV grid faults

This paper analyses the impact of symmetrical and unsymmetrical voltage sags caused by distribution and transmission faults on the dynamic behaviour of small salient-pole synchronous machines connected to the MV grid, and proposes the modification of the protection settings to increase machine availability. As faults are considered, the voltage recovery is assumed to occur in a discrete way, in the fault current zeroes. The voltage sag effects include current and torque peaks, speed and load angle increase, and possible machine instability. Results from many cases are discussed, especially the positive-sequence voltage influence on machine stability. The study concludes that machine availability during unsymmetrical sags can be increased by modifying the undervoltage and overcurrent protection settings (the former should be based on the positive-sequence voltage instead of line-to-line voltages used in current settings).

A data pipeline for PHM data-driven analytics in large-scale smart manufacturing facilities


 
 
 The term smart manufacturing refers to a future-state of manufacturing, where the real-time transmission and processing of information across the factory will be used to produce advanced manufacturing intelligence that can optimize every aspect of its operation. In recent years, initiatives and groups such as the Smart Manufacturing Leadership Coalition (SMLC), Industry 4.0, and the Industrial Internet Consortium (IIC), have led the way by bringing together industry, academia and government to establish policies, roadmaps and platforms to support smart manufacturing. Although there are many characteristics that can be associated with smart manufacturing across these initiatives, a common theme is the emphasis on transitioning operations from reactive and responsive, to predictive and preventative. The research presented in this paper focuses on the development of a data pipeline that supports the development of data-driven Prognostics and Health Management (PHM) applications. In the context of smart manufacturing, PHM enables facilities to transition from preventative and reactive maintenance strategies, to predictive, preventative and condition-based strategies. The benefits that can be derived by PHM are aligned with those of smart manufacturing, which include the opportunity to decrease costs, increase machine availability, reduce energy consumption, and improve production yield.
 However, the process of ingesting, cleaning and transforming real-time data streams for data-driven PHM is a difficult, complex and time-consuming task, with estimates from business intelligence projects ranging from 80% to 90% of total project effort. This effort may only be exacerbated further in manufacturing environments due to additional technology challenges, such as low levels of standardization, disparate protocols and interfaces, and ad hoc data management. While emerging technologies such as Cyber Physical Systems (CPS) and Internet of Things (IoT) can overcome many of these challenges and provide an open platform for transmitting data, existing large-scale manufacturing facilities that are subject to compliance, regulation, and stringent quality
 
 
 assurance policies may not be able to adopt these technologies in the short-term due to the associated cost, risk and effort. Therefore, PHM applications that need to access data streams in large-scale manufacturing facilities must do so using transparent data integration that does not discriminate between emerging and legacy technologies in the factory. To this end, this research presents a real- time, scalable, robust, and fault tolerant data pipeline for ingesting, cleaning, transforming, processing and contextualizing time-series data from a wide-range of sources in the factory.
 
 

A Bayesian Inventory Model Using Real‐Time Condition Monitoring Information

Lack of coordination between machinery fault diagnosis and inventory management for spare parts can lead to increased inventory costs and disruptions in production activity. We develop a framework for incorporating real‐time condition monitoring information into inventory decisions for spare parts. We consider a manufacturer who periodically replenishes inventory for a machine part that is subject to deterioration. The deterioration is captured via condition monitoring and modeled using a Wiener process. The resulting degradation model is used to derive the life distribution of a functioning part and to estimate the demand distribution for spare parts. This estimation is periodically updated, in a Bayesian manner, as additional information on part deterioration is obtained. We develop an inventory model that incorporates this updated demand distribution and demonstrate that a dynamic base‐stock policy, in which the optimal base‐stock level is a function of some subset of the observed condition monitoring information, is optimal. We propose a myopic critical fractile policy that captures the essence of the optimal policy, but is easier to compute. Computational experiments indicate that this heuristic performs quite well relative to the optimal policy. Adaptive inventory policies such as these can help manufacturers to increase machine availability and reduce inventory costs.

Fault detection and diagnosis of an industrial steam turbine using fusion of SVM (support vector machine) and ANFIS (adaptive neuro-fuzzy inference system) classifiers

The subject of FDD (fault detection and diagnosis) has gained widespread industrial interest in machine condition monitoring applications. This is mainly due to the potential advantage to be achieved from reduced maintenance costs, improved productivity and increased machine availability. This paper presents a new FDD scheme for condition machinery of an industrial steam turbine using a data fusion methodology. Fusion of a SVM (support vector machine) classifier with an ANFIS (adaptive neuro-fuzzy inference system) classifier, integrated into a common framework, is utilized to enhance the fault detection and diagnostic tasks. For this purpose, a multi-attribute data is fused into aggregated values of a single attribute by OWA (ordered weighted averaging) operators. The simulation studies indicate that the resulting fusion-based scheme outperforms the individual SVM and ANFIS systems to detect and diagnose incipient steam turbine faults.

Transformation of mechanized locomotives into semi-autonomous machines

Computerized communications (radio) command, control, and monitoring (C/sup 4/M) technology has been developed to replace mechanical controllers on 35 ton locomotives (loci). The C/sup 4/M technology uses medium frequency (MF) underground radio technology developed in prior US Bureau of Mines research to achieve high-quality voice and data transmission in the mining complex. Robust electrical conductors such as rail, air/hydro pipe, and cable provide a signal transmission medium with an attenuation rate of 4 dB/304 m (1000 ft). The signal-to-noise ratio (SNR) exceeds 30 dB in the transmission system. A surface supervisory computer communicates with the C/sup 4/M controller via MF radio to create a semiautonomous locomotive. Diagnostic algorithms monitor locomotive sensors and locate failures in the propulsion system. The C/sup 4/M controller has reduced propulsion system diagnostic time from an average of 40 to less than 10 min. The increased machine availability time contributes 17000 tons of additional ore to the trains' haul age potential each year. >