Högskolan i Skövde

Transitioning from fossil fuels to sustainable energy sources, however necessary, will increase energy consumption, representing a significant challenge to grid capacity. One such area is the manufacturing of cast iron components for the automotive industry. Traditionally, these melting furnaces have been powered by fossil fuels, e.g., coke, producing a high carbon footprint. Replacing those furnaces with electric melting furnaces powered by sustainably sourced energy will reduce the CO2 footprint but increase the electrical power peaks during the melting cycle. Inevitably, this transition will also affect the local grid and the communities where the manufacturing plants are located.

This paper will present a proof-of-concept decision-support system powered by several predictive models on a cloud platform. The melting cycle in the melting furnace will produce a large electrical power peak during operation, which could coincide with other power peaks on the local grid. The aim is to shift the start of the melting cycle to maintain production output while minimizing the coinciding power peaks from a local grid standpoint.

To achieve this, three predictions, or forecasts, are utilized. First is a prediction on future plant energy consumption, second is a prediction on local grid power consumption, and third is a prediction of future molten iron need. The energy consumption predictors utilize machine learning, while the molten iron need predictor utilizes a digital twin of the production environment. The three predictions are used to create an optimized melting schedule for the next eight hours, which is repeated every thirty minutes to react to changes in the predictors. Reducing peak power consumption brings benefits for individual companies in addition to the communities they operate in. By providing decision support for operators, they can make decisions based on data to reduce the load on the energy grid, for value both to industry and society. For future work, the proof of concept will be expanded to encompass more melting furnaces and additional production lines and allow the operators more control over the digital twin through the decision support system.

Wood supply chains are complex, involving many stakeholders, intricate processes, and logistical challenges to ensure the timely and accurate delivery of wood products to customers. Weather-related variations in forest road accessibility further complicate operations. This paper explores the challenges faced by forest managers in targeting many delivery requirements—four or more. To address this, simulation-based optimization, using NSGA-III, a many-objective optimization algorithm, is proposed to simultaneously optimize often conflicting objectives primarily by minimizing delivery lead time, delivery deviations in backlogs, and delivery variation. NSGA-III enables the exploration of a diverse set of Pareto-optimal solutions that show trade-offs across a flexible set of four, or more, delivery objectives. A Discrete Event Simulation model is integrated to evaluate objectives in a complex wood supply chain. The implementation of NSGA-III within the framework allows forestry decision-makers to navigate between different harvest schedules and evaluate how they target a set of preference-based delivery objectives. The simulation can also provide detailed insights into how a specific harvest schedule affects the supply chain when post-processing possible solutions, facilitating decision making. This study shows that NSGA-III could substitute NSGA-II to optimize the wood supply chain for more than three objective functions.

Manufacturing companies are gradually moving from Industry 4.0’s technology focus to Industry 5.0’s sustainability focus, and identifying and addressing energy bottlenecks is a part of this transition. In practice, this is challenging due to limited availability of energy data and its poor integration with systems like MES and SCADA. Energy dashboards are capable of consolidating energy data, visualizing consumption patterns, and tracking related KPIs for sustainability. However, most existing implementations are limited to facility-level overviews or machine-specific views without consideration of operational details. To identify energy bottlenecks, the dashboards must also analyze machine states, batch sizes, product mixes, and cycle times. Therefore, this paper presents a Python-based web application built with the Dash framework and open-source packages. The application integrates data from EMS, MES, and SCADA systems. It is capable of performing statistical time-series analysis, joint energy-stop analysis, state-based mapping of energy use, and visualizing various Key Performance Indicators. The proposed integrated dashboard targets discrete manufacturing and is demonstrated on a gear machining line at Volvo Group Trucks Operations. The dashboard currently operates offline with data from enterprise systems, but aims for real-time API integration as a digital twin in the future. This could support simulation for detecting inefficiencies, predicting energy bottlenecks, and optimizing energy consumption.

As manufacturing systems become increasingly complex and data-rich, there is an opportunity to identify predictive patterns in the data, including deeper patterns and more general, broad principles for effective use of our production infrastructure. Simulation-based multi-objective optimization provides additional options and predictions for industrial decision-makers; however, traditional analytics approaches struggle to provide the necessary insights. More sophisticated methods, such as data mining, provide greater insight; however, they require technically proficient users to achieve results. This paper will examine an example application in which data mining methods were applied to the results of multi-objective optimization and modeled as a knowledge graph. Our study presents the development and evaluation of an LLM-based Graph Retrieval-Augmented Generation (GraphRAG) tool that can translate natural language queries into Neo4j graph database queries for manufacturing data analysis. We present both successful query generations and identify failure modes, providing insights into the current capabilities and limitations of this approach. The paper includes a detailed use case description, documenting specific manufacturing analytics requirements and the corresponding graph query patterns needed to extract meaningful insights.

Human–robot collaboration solutions have gradually become popular, in which some tasks are performed by robots and others by humans. Designing such a production cell requires simultaneous consideration of human-centered factors, machine-focused mechanical design, and system engineering in the early planning stages. However, different objectives often conflict (e.g., speeding up a robot can improve productivity while compromising energy efficiency), and the same variables can affect multiple models and simulations simultaneously (e.g., a machine where humans and robots collaborate can influence both the operator’s working posture and the robot’s cycle time). Therefore, multidisciplinary tools and multi-level optimization are needed to model, simulate, and optimize elements such as production flows, robotics, and human operators to balance objectives related to cycle time, energy consumption, and worker well-being. In this paper, we formulate an approach that integrates different simulation tools and a bi-level optimization framework to balance worker well-being, cycle time, and energy consumption. We demonstrate this approach through a real industrial case of designing a work cell for elevator pipe assembly in a grain conveying system, where ABB RobotStudio is used for robotic simulation and IPS IMMA for human simulation. IBM ILOG CPLEX Optimization Studio is employed for the top-level task allocation optimization, and a set of results is presented based on data extracted from the lower-level robot-centered optimization. The results show that our approach can effectively balance different objectives by incorporating detailed information from different levels of the work cell design.

Harvest scheduling and transport are crucial for the delivery performance of a wood supply chain, ensuring that product volumes are delivered on time and in the right quality. This paper suggests three delivery performance objectives for the wood supply chain: service level, lead time, and throughput. It presents a framework for optimizing these objectives by finding trade-off solutions using simulation-based multi-objective optimization. Due to the complexity of the wood supply chain, discrete-event simulation is used to evaluate delivery performance from harvesting to customer delivery. The harvest scheduling problem is formulated as a permutation optimization solved by a customized NSGA-II algorithm with a comparison of three crossover mechanisms implemented: Random Key Simulated Binary Crossover, Order Crossover, and Partially Mapped Crossover, specifically designed for general forestry permutation optimization problems. Analyzed with a heatmap for the visualization of the mapping of the decision space to the Pareto-optimal solutions, the results indicate that the Partially Mapped Crossover performs best. Other simulation-optimization generated data are processed and visualized in an interactive, web-based dashboard for decision-makers, such as forest managers, allowing them to analyze meta-heuristically optimized solutions in both the solution and decision spaces, guiding them to find the most suitable harvest schedules. 

With the increase in complexity of industrial systems it becomes more and more challenging to make well-grounded decisions for system design and operation. Following the concept of Virtual Factories with Knowledge-Driven Optimization (VF-KDO), this paper proposes a graph database approach to support knowledge-driven and simulation-based optimization. With the mapping of a VF-KDO ontology to a graph database, competency questions that facilitate traceability, transparency, and group decision making can be answered. This is exemplified with an industrial use case and a scenario form academic education.

Understanding delay conditions and making accurate predictions are essential for optimizing turnaround and taxi times, which in turn reduces fuel consumption and lowers CO2 emissions in airport operations. However, while existing research has explored the impact of various prediction models on airport operations, it often overlooks the performance of Collaborative Decision Making (CDM) variables when discussing delay conditions. The implementation of CDM at major European airports has led to a milestone-based approach within airport operations, particularly in the turnaround operations, segmenting these operations with unique features. The purpose of this paper is to systematically investigate the efficacy of various machine learning techniques, such as linear regression, regression trees, random forests, elastic nets, and multi-layer perceptrons (MLP), in accurately predicting delay categories within the CDM framework. For this purpose, we analyzed CDM operational data from Madrid Airport, with at least 166,185 flight observations. Our findings illustrate a training methodology on how different models vary in prediction accuracy when applied to CDM operational data. We applied the SHAP (SHapley Additive exPlanations) method for feature importance analysis of all our independent variables to interpret the output of our machine learning models. Our results indicate that linear regression and elastic nets are the most effective machine learning models for achieving high prediction accuracy within the CDM framework. To test their robustness, we extended the analysis with predictions for better schedule times for taxi times on arrival and depature for selected runways using a different dataset. Our results contribute by showcasing a training methodology, highlighting how elastic net model as the best-performing model can be adopted for turnaround operations. In conclusion, we discuss the implications of our results for runway demand policies and use of airport resources such as gate & runaway allocation. 

Virtual manufacturing, simulation, and optimization provide a wealth of knowledge about the possibilities of future production systems so as to support decision makers. However, this knowledge usually remains with a handful of domain experts, is not captured and is hard to share even within the same team. At the same time, simulations can benefit from the incorporation of linked data from real factories once a process is running. Graph databases provide a possible approach to storing and managing this form of interrelated heterogeneous data, with powerful querying capabilities that can identify important or interesting patterns that might otherwise remain hidden. Current research focuses on one or two aspects of this problem but does not address all at once, despite the potential benefits of the combination. This paper provides a broad literature review of the current directions within research with a special focus on how graphs can support finding knowledge within Virtual Factories, used by larger teams for industrial planning and optimization.

Batching and scheduling orders for ovens in manufacturing is a typical combinatorial optimization problem, and it is critical for production efficiency and customer satisfaction. Customized settings and randomness happened in practical production including the maximum waiting time of an order, capacity thresholds, and availabilities of ovens make the problem more complex. In this paper, we build a simulation model in ten ovens for a heat-treatment process line of cutting tools according to real data from an industrial case, and the model embeds a detailed control logic that integrates existed scheduling methods with various dispatching rules and parameters. After determining an optimal number of running ovens in the production line by simulation-based optimization, we then propose a multi-objective optimization (MOO) enhanced deep reinforcement learning (DRL) approach to schedule orders for the ovens. The DRL agent learns to use an appropriate dispatching rule at a scheduling time point to select orders and formulate a batch. Along with reducing the average tardiness of the orders and maximizing the average effective utilization of the ovens, we find that the explored MOO enhanced DRL approach is more flexible and effective than the heuristic method.