Industrial Engineering and Applications – Europe

This study discusses how to optimize the cost for condition-based maintenance, specifically in the case of a two-component machine system. This is important because, in manufacturing systems that are highly dependent on machines, the total cost of maintaining stochastically degrading components can be high (i.e., as a result of inspections, general repair including replacement, and also the cost of lost production and process rejects due to machine breakdowns). This study hypothesizes that the minimum total cost of a condition-based maintenance policy for a two-component series system can be found by jointly solving for (a) the optimal inspection interval, (b) the general repair conditions for the two components, and (c) the cost estimation of process rejects and breakdown during every maintenance inspection and maintenance performed; doing all of these will give a realistic cost structure as encountered in real-life manufacturing systems. Given this hypothesis, this study models the two-component system by setting up the operating states and the breakdown states and defining the system variables. Then, this study establishes the decision variables, i.e., for the inspection state and the maintenance states, and thereafter performs the exhaustive Monte Carlo simulation approach to get the optimal variables. Finally, this study uses the optimum decision variables to come up with policies for condition-based maintenance. To prove the validity of this hypothesis, a case study is performed on a two-component process of a cutter-grinder assembly of a local fast-moving consumer goods company in the Philippines. By applying the approach proposed by this study that integrates stochastic degradation, frequency of inspections, process rejects and general repair scenarios, the total cost of annual maintenance is seen to be reduced by 44%.

Maintenance within the mining industry is a global challenge, demanding innovative solutions due to its multifaceted nature. The complexity, size, competition, escalating costs, and the need for a proficient workforce present formidable obstacles. This study focuses on optimizing technically skilled worker allocation in the mining industry’s maintenance department, developing a continuous-event simulation model. The ever-increasing intricacy of mining equipment necessitates heightened reliability to ensure efficient, continuous production. Adequate maintenance is crucial in this context. The research problem revolves around the critical role of technically skilled workers during maintenance operations, where their absence can lead to delays and operational inefficiencies. Skilled maintenance technicians are indispensable, capable of accurately diagnosing equipment issues, reducing costly breakdowns, and minimizing downtime through preventive maintenance. They also expedite the repair process when required. To address these challenges, our study introduces a continuous event simulation algorithm designed to minimize costs. This algorithm takes into account factors such as production losses and maintenance workforce expenses while maximizing equipment utilization. By doing so, the research contributes to the field by emphasizing the importance of a skilled workforce in mining maintenance, ensuring equipment longevity, performance, and safety. The contribution of this study lies in its practical application of advanced algorithms to optimize technically skilled worker allocation, mitigating operational challenges and highlighting the crucial role of skilled maintenance management in the evolving landscape of mining equipment.

Increasing digitization and automation are currently bringing a revolution to many processes and sectors of the economy. The industry of production in particular is undergoing a fundamental transformation through the implementation of digital and networked system elements. In production control, traditionally based on established methods and experience, advanced technologies such as artificial intelligence are opening up entirely new options. Artificial intelligence’s ability to analyze large amounts of data in real time and make instant decisions from it brings fresh perspectives to production control. Until now, there have been only a few surveys that examine the current status as well as future requirements for production control. Our research focuses on the current state of IT systems used to support decision-makers in production control. It also sheds light on the requirements needed to improve these systems, especially in light of advancing technological developments and increasing digitalization. The article is dedicated to the results and provides an overview of future expectations.

The production of automotive wire harnesses now requires a significant amount of manual labor. Even so, a greater level of automation is required due to present and future application demands such as the miniaturization of electronic components, the monitoring of process parameters, the growing need for paperwork for processes, and the rise in payments. Technology helps manufacturing organizations in the most important aspects of the design-to-manufacturing process. This is relevant to the wire harness sector, which is a crucial part of industrial automotive manufacturing. The automotive wiring harness suppliers designed Computer-Aided Design technologies (CAD) to support wire harness assembly operations and design work. Even with the application of these techniques, engineers will still have to do trial-and-error work to find effective assembly techniques. This study presents a new approach to optimize the wire harness assembly procedures without focusing on trial and error or the experience of experienced engineers to develop operational assembly processes. The most crucial and challenging step in the assembly process sequence is taping routed cables. The taping process’s complexity is mostly determined by how the jig is set up on the workstation and the tape method. As a result, the suggested technique models and optimizes the tape direction and jig arrangement.

Occupational accidents have always been very important for all industries due to its significant impacts on projects and society. Whereas the case in construction industry is also similar due to higher number of occupational accidents recorded in various countries. In current dynamic technological era, conventional Health and Safety (H&S) practices are not sufficient thus new approaches are important to explore. It is observed from the detailed literature review that there are various existing tools designed by various researcher in different scenario but there is a lack to device an integrated approach. Therefore, this paper proposes an integrated approach to monitor the H&S at site. The proposed framework will help the decision makers to manage and monitor the health and safety practices at site in a real time environment.

Path selection in multistage manufacturing process has a considerable impact on product quality because uniform product quality cannot be ensured across all equipment. Accordingly, we present a method for choosing the best path while taking product quality into account. In this paper, we firstly introduce a quality index and categorize product grades. Building upon the criteria, we also propose an approach based on path scores to identify machine sequence paths that can improve product quality. To validate its efficacy, we apply the approach to the industry-like data, successfully pinpointing both the most and least favorable paths.

Continuous increase in demand for land transport calls for motorists’ heightened usage of roads. This interconnectivity of road infrastructures features the various types of road intersections seen worldwide. When constructed, roads may be made functional with or without traffic lights. This study will feature a traffic management system for a signalized t-intersection to mitigate vehicular traffic and road congestion. A simple single-objective optimization model was developed to minimize the traffic system’s overall cycle time and determine the green, red, and amber timing using the model and a traffic simulation to observe vehicle density, delay, and volume trends. From an unsignalized base model, three (3) signalized simulations were conducted with varying road configurations and phase plans to address conflict points in a t-intersection. A significantly efficient trend in conflict point reduction and phase planning improvement was achieved, meeting the objective of minimizing the cycle time.

During the steering manoeuvre, due to the variations such as vehicle parameters, speed and tire grip, vehicles may undergo understeer or oversteer, which may lead to vehicle deviation from expected path and lose of control, greatly threatening passengers’ safety. Therefore, this paper proposes a coordinated control method for the lateral stability of four wheel independent steering vehicles with the surrounding condition considered, which could be adaptable to the surrounding environment and has excellent control performance. Firstly, a four independent wheel vehicle model combining longitudinal and lateral force is established. Secondly, a LQR controller with its parameters adjustable according to the surrounding condition is designed. Finally, the effectiveness of the control algorithm is verified via simulation.

Scheduling classes is an essential and challenging task due to academic rules and constraints imposed by the system. Previous studies have tackled the problem of class scheduling, but teacher schedule and course preference have not been incorporated. The usability of previous software developed were also not established. This study aims to create an automated tool that will minimize the time and errors the department heads encounter in creating a faculty schedule. The Interactive Scheduling Heuristic-based Application (ISHA) was developed using a user-centered design approach. Seventeen department administrators were interviewed to determine the tasks involved in the scheduling process, the constraints, and best practices. The contents of the software were determined from the outcome of a task analysis. The system was designed to be embedded in the web application and the algorithm itself when autogenerating a schedule. The algorithm prevents schedules from overlapping with one another. The usability of the application was assessed using efficiency by comparing the speed of scheduling with and without the tool. ISHA was able to ease the burden of the department heads in preparing a schedule that considers all the constraints imposed by the system. The scheduling task had been shortened, less prone to errors, and had been more oriented with the capabilities of the department heads.

Continuous manufacturing of pharmaceutical tablets integrates multiple unit operations such as twin screw granulation and fluidized bed drying to transform powder into final dosage form such as tablets. However, complex process interactions can lead to variability in critical quality attributes including moisture content of the produced granules. This study presents an innovative multi-stage modelling framework to predict granule moisture content based on the twin screw granulator and the fluidized bed dryer process parameters. Machine learning techniques, including gradient boosting regression, and support vector regression were utilised to enhance predictive performance in ensemble method. Using data from a pilot-scale integrated continuous line, the stacking ensemble model achieved excellent accuracy with (\(R^2\)) of 91% for moisture content prediction. The Machine learning modelling framework demonstrates strong potential for advancing process knowledge, and optimization in continuous manufacturing of pharmaceutical tablets based on wet granultion.

The rise of mass-individualization has underscored the significance of Mixed-Model Assembly Lines (MMALs) for producing diverse products on the same line. The Car Sequencing Problem (CSP) tackles short-term balancing in an MMAL by emphasizing the use of spacing rules to manage the space between each pair of work-intensive products that possess specific characteristics. In this study, we tackle two challenges within the CSP context. The first challenge involves exploring CSP with cross-ratio constraints that takes into account the dependency between different characteristics. As the second challenge, we study the CSP under two states where spacing rule violations are not allowed (hard) and allowed (soft). We develop two constraint programming models for the mentioned states and evaluate the performance of the models using several real-world assembly lines’ instances. The findings enhance understanding of each model’s strengths and weaknesses. Given the inherent complexity of real-world problems, the soft model may find more practical and effective application. This research enriches the realm of problem-solving in MMALs by offering valuable insights and introducing the main challenges in the CSP.

We consider an order-picking system for a warehouse divided into corridors with two-layer shelves being arranged in the shape of a U in each corridor. Given an order in a corridor, the focus is on the optimization of the picking sequence and on locating the movable depot in the most convenient location. Two iterative algorithms based on constraint programming are proposed. Computational experiments position the new methods in the existing literature, showing that they are operatively effective.

We also show how allowing the depot to be allocated away from the central axis of the corridor can lead to substantial time savings, especially for small orders. This strategic option had not been considered in the previous literature, but can be easily implemented in modern warehouses.

In the last decade industrial systems have been affected by increased challenges, and among those the need for more sustainable production and logistics has been called into question by both practitioners and academia. Within industrial networks, logistics facilities traditionally represent key nodes as they have a direct impact on both companies’ service levels and logistics costs. Recently, their complexity has dramatically increased due to their ever-demanding requirements and pressures from stakeholders and society. In this context, companies have started to search for solutions for greener warehousing processes and energy efficiency improvements. Still, on the academic side, a limited number of studies have been found addressing the quantification of logistics facilities’ environmental performance, the impact of the green warehousing practices in place, and the related effects on warehouse consumption and emission reduction. This paper aims to address this research gap by proposing a simulation-based approach where multiple scenarios of a real logistics facility are discussed, grounded on a conceptual framework that offers a roadmap towards sustainable and energy-efficient warehousing. Different scenarios are outlined, and the related performances are examined in terms of energy consumption and CO2_eq emissions. Implications of the results are discussed and streams for future investigation are identified.

Nonnegative tensor factorization (NTF) plays a crucial role in extracting latent factors and predicting future sales from purchase data consisting of user and item attributes. However, the increase in these attributes leads to tensor data becoming sparse, causing a reduction in decomposition accuracy. For example, when there are numerous combinations of unavailable item genres and prices, the purchase history data becomes sparse and follows a distribution where all its elements are zero. To address this issue, we propose a novel NTF method assuming zero-inflated Poisson (ZIP) distribution based on Expectation-Maximization (EM) algorithm. This enables us to effectively handle sparsity in high-dimensional multiway data and identify combinations of user and item attributes that are potentially not likely to be purchased. We verified the effectiveness of the proposed approach through numerical experiments using real-world e-commerce data. The results showed our proposed ZIP model outperforms existing methods in both in-sample and out-of-sample experiments. Moreover, the proposed method qualitatively demonstrated the effectiveness of handling sparsity.

In the automotive service and spare parts distribution sector, effective supply chain management is paramount as it directly impacts customer satisfaction, profits, and overall competitiveness. To optimize these crucial aspects, a robust fuzzy logic framework has been developed, taking into account the intricate connections between customer demand, customer sentiment, product availability, and pricing strategy. This study focuses on after-sales services provided by Moroccan automobile companies, utilizing workshop entry data from Enterprise Resource Planning (ERP) systems across multiple cities in the country.

To accurately capture customer sentiment, a Bidirectional Long Short-Term Memory (BiLSTM) with attention model, a deep learning strategy, is employed to complete any missing customer sentiment data. Leveraging the power of neural networks, this approach effectively analyzes and predicts sentiment patterns. Once the sentiment data is completed, a Fuzzy Logic model is utilized to determine the most optimal pricing strategy. By considering various factors through fuzzy sets and rules, this model allows informed decision-making to strike the right balance between profitability and customer satisfaction.

The employed models demonstrated strong performance, as evidenced by metrics such as MSE and R2. Specifically, the BiLSTM-attention model achieved an R2 of 0.87, while the fuzzy logic model registered an impressive R2 of 0.91. Furthermore, when juxtaposed with alternative models, our framework’s superior efficacy became evident.

As an essential steel product, wire rods have specific requirements regarding their physical properties. Especially for wire rods for automotive springs, it is important to ensure consistent hardness throughout the product. Because traditional hardness testing methods are destructive and sample-based, they have the potential to overlook the non-uniformity of wire rod hardness. This paper presents the application of a convolutional neural network (CNN) to thermal imaging to address these issues. The model segments the thermal image of a wire rod after cooling, separating the temperature of the wire rod and the background on a pixel-by-pixel basis. This temperature data is used to calculate the cooling rate and helps to predict the hardness of the wire rod along its entire length. Experimental results show that the U-Net-based model outperforms a simple FCN model in the segmentation task. This approach provides a more comprehensive quality inspection of wire rod, bringing both economic and quality benefits to the steel industry.

The increasing use of synthetic training data has emerged as a promising solution in various domains due to its ability to provide accurately labelled datasets at a lower cost compared to manually annotated real-world data. In this study, we investigate the use of synthetic data for training deep learning models in the field of industrial object recognition. Our goal is to evaluate the performance of different models trained with varying ratios of real and synthetic data, with the aim of identifying the optimal ratio that yields superior results. In addition, we investigate the impact of introducing randomisation into the synthetic data on the overall performance of the trained models. The results of our research contribute to the understanding of the role of synthetic data in industrial object detection.

In this study, we propose a framework for detecting, classifying, and visualizing unknown patterns in semiconductor wafer defect analysis to improve automation in the field. Rapid advancements in semiconductor processes and equipment have led to the emergence of new defect types, most of which are analyzed and identified based on engineers’ experience and judgment. Current approaches struggle with limited labeling, emerging defects, and class imbalance, and although pattern recognition and deep learning techniques have been applied in research, they do not provide a complete solution. We present a method that can quickly detect various emerging defect patterns and ensure high classification accuracy for known defect types. To achieve this, we utilize One Class SVM and Transfer Learning-based ResNet50 backbone, which can be easily implemented on-site. The proposed method uses the one-class SVM method and the validation threshold of each classifier to perform multi-stage unknown defect pattern detection. This approach overcomes the limitations of traditional defect analysis, supporting the identification of new defect types and enhancing engineers’ work efficiency. Furthermore, we employ T-SNE and DBSCAN techniques for dimensionality reduction and visualization, providing high accuracy and dimensionality reduction in identifying new defect patterns. These techniques aid engineers in timely labeling and decision-making, ensuring a more efficient response to emerging defects in the semiconductor industry. Consequently, this study offers a comprehensive framework that addresses the challenges of limited labeling, emerging defects, ultimately improving the performance of semiconductor wafer defect analysis. The effectiveness of the proposed model is evaluated through various experiments.

This paper falls within the industry 4.0 and tackles the challenging issue of maintaining the Digital Twin of a manufacturing warehouse up-to-date by detecting industrial objects and estimating their pose in 3D, based on the perception capabilities of the robots moving all along the physical environment. Deep learning approaches are interesting alternatives and offer relevant performances in object detection and pose estimation. However, they meet the requirement of large-scale annotated datasets for training the models. In the industrial and manufacturing sectors, these massive datasets do not exist or are too specific to particular use-cases. An alternative aims to use 3D rendering software to build annotated large-scale synthetic datasets. In this paper, we propose a framework and guidelines for creating synthetic datasets based on Unity, which allows the 3D-2D automatic object labeling. Then, we benchmark several different datasets, from planar uniform background to 3D contextualized Digital Twin environment with or without occlusions, for the industrial cardboard box detection and 6D pose estimation based on the YOLO-6D architecture. Two major results arise from this benchmark: the first underlines the importance of training the deep neural network with a contextualized dataset according to the targeted use-cases to achieve relevant performances; the second highlights that integrating cardboard box occlusions in the dataset tends to degrade the performances of the deep-neural network.