Constant advancements in biomanufacturing require operational frameworks that enable accurate comparisons between similar production processes for performance and cost optimization purposes. However, the extensive and heterogeneous nature of available data which is often presented in different formats, characterized in different ways, and reported in disparate units - poses a significant challenge to an efficient and effective comparative analysis. In this paper, the potential benefits of applying the biomanufacturing production process ontology to provide a common ground for addressing data heterogeneity issues are investigated in the context of process design optimization with respect to key performance indicators (KPIs). Using acetic acid fermentation as a case study, three semi-continuous processes involving different bacterial species and feedstock sources were analyzed. Existing industrial ontologies were utilized, and additional ontologies were developed to facilitate the normalization of notions and link data from multiple domains, including life cycle analysis, process, production, and materials. Following ontology development best practices, competency questions and respective queries were formulated to validate the ontology, with the aim to evaluate KPIs, including the material intensity and the number of fermentation cycles each microorganism can sustain. Results demonstrate that the ontology effectively resolved data inconsistencies, enabling more accurate comparisons and providing deeper insights into the differences in processes across various strains and conditions.
To remain competitive in the long term and sustain or expand market share, companies rely on continuous innovation. In practice, investment decisions are typically based on expected economic value, although such evaluations are challenging in early innovation phases due to limited data availability. In particular, the future economic potential of the results from fundamental research is difficult to predict. The Collaborative Research Center 1368 investigates the influence of an oxygen-free atmosphere on the active zones in production processes as an innovation to develop future-proof production technologies. One example is the compound casting of oxide-free aluminum-copper composites with fullsurface metallurgical bonds, which significantly improves thermal conductivity compared to conventional joining methods. A relevant application lies in computer components such as graphical processing units, where power electronics generate substantial heat loss and demand efficient, sustainable cooling solutions. These require improved material performance to reduce thermal resistance and enable more effective heat dissipation, potentially extending the product use phase. While the innovation introduces additional manufacturing effort, it also offers potential long-term economic benefits during the use phase. This paper proposes an approach to integrate use phase benefits into early-stage innovation evaluation, capturing economic effects that extend beyond the production stage. This study supports a more holistic and forward-looking assessment of innovation potential by focusing on functional performance improvements and their relevance for customer value, particularly through the extended product use phase. The findings can be fed back into the development process, helping identify and improve key value drivers early on.
This study examines how inherent contradictions affect supply chain planning (SCP) practices in engineer-to-order (ETO) settings through a multi-case study of five companies. The research identifies sixteen contradictions across four paradox categories: performing, organizing, belonging, and learning paradoxes. The findings reveal that contradictions shape SCP practices while SCP practices simultaneously influence how contradictions are managed. Companies employ four management strategies: acceptance, spatial separation, temporal separation, and synthesis to handle contradictions. One of the approaches acknowledge contradictions as inherent features of ETO settings rather than a problem to eliminate. Companies with structured SCP processes demonstrate better contradiction management capabilities, creating virtuous cycles of continuous improvement. The research contributes to paradox theory by validating the dynamic equilibrium model in SCP contexts and identifying organizational capabilities that enable effective paradox management. For practitioners, the study suggests that explicitly acknowledging contradictions, aligning organizational structures, and developing balanced performance metrics can transform competing demands into sources of competitive advantage rather than operational constraints.
Norwegian yards face difficulties in operational coordination and the adoption of effective technical solutions, reducing their overall productivity. To remain competitive and fully leverage the vast amount of data, it is essential to adopt innovative technologies and methods, such as digital twins and artificial intelligence. This study aims to map the current applications of digital twin technology in yards, exploring its potential, challenges, and needs. A comprehensive literature review and a case study from a Norwegian yard are conducted. Research reveals that applications are often limited to small or specific parts of the yard, rather than full-scale operations. While the use of 3D visualization tools and artificial intelligence optimization methods is widespread, the findings highlight the need to implement digital twins and artificial intelligence technologies more extensively in production and across entire yard operations to enhance overall efficiency and productivity.
Digital transformation (DT) has revolutionized manufacturing by enhancing efficiency, productivity, and competitiveness. However, adoption remains concentrated among large corporations, while microenterprises (MEs) often lag due to limited resources, which hinders their growth. Conventional DT strategies focused on SMEs are ill-suited for MEs, which face distinct challenges such as informal structures, minimal capital, low digital literacy, and reliance on analog machinery-risking a widening digital divide. This study investigates DT in the context of manufacturing MEs, focusing on migrant-run enterprises in Oman operating under severe constraints. Based on a qualitative methodologyincluding interviews, field observations, and participatory work-it proposes a frugal, two-layer DT framework tailored to the socio-technical realities of MEs. The digital layer emphasizes immediate digitalization of business processes using accessible tools like spreadsheet dashboards and pricing templates, co-designed with ME stakeholders to ensure relevance and usability. The physical layer enables access to digitally capable production infrastructure through low-cost, DIY opensource machine tools (OSMT). This is demonstrated through the local replication of an open-source 3D printer design. By prioritizing affordability and leveraging open-source technologies, the proposed approach reduces dependency on proprietary systems and offers a practical path toward inclusive digital transformation. This research contributes to emerging discussions on DT for MEs, offering actionable strategies to enhance their competitiveness in an increasingly digital manufacturing landscape.
The design of large-scale equipment manufacturing systems plays a crucial role in ensuring product performance, optimizing production efficiency, and reducing lifecycle costs. Effective reuse of domain knowledge is essential for maintaining both the quality and efficiency of manufacturing system design. Although existing knowledge graph technologies standardize the representation and storage of such domain knowledge, the complex design constraints and multiple optimization objectives of manufacturing systems still pose significant challenges to the efficient reuse of domain knowledge. Recent advancements in the large language model (LLM) and retrieval-augmented generation (RAG) have led to the emergence of graph retrieval-augmented generation (GraphRAG), which presents a promising approach to overcoming these challenges. This paper proposes a novel GraphRAG-based method, Design-on-Graph, to support knowledge management and automated generation of design plans for manufacturing systems. This method employs the LLM to intelligently retrieve and verbalize structured domain knowledge through multi-turn conversations, achieving highefficiency knowledge management for manufacturing systems. Additionally, the retrieved domain knowledge is systematically archived within conversation history, providing contextual support for LLM-driven reasoning tasks to streamline automated design processes. Finally, a case study on an aircraft fuselage joint system is conducted, and an AI agent incorporating all the above functionalities is developed to demonstrate and evaluate the performance of the proposed Design-on-Graph.
The advancement of digital technologies and the consolidation of the platform economy have driven new models of logistics organization, with crowdsourcing standing out as an innovative alternative for the last mile. This model mobilizes underutilized resources from the population – such as vehicles and informal labor – through digital platforms that connect service providers to users. In large urban centers in Brazil, such as São Paulo and Rio de Janeiro, these solutions have changed traditional distribution paradigms. This article aims to evaluate the performance of the Loggi, Rappi, and 99Frete platforms, considering three dimensions: operational efficiency, economic viability, and social and regulatory impacts. The methodology is based on international studies and the concept of crowd logistics, identifies advances in flexibility and inclusion, and also points out challenges such as job insecurity, urban sustainability, and legal gaps. The study contributes to the debate on innovation, social justice, and digital transformation in Brazilian urban logistics.
Although cooperative robots have been rapidly introduced in warehouses, high turnover remains a major problem in semi-automated warehouses. Order picking is a representative example of a process where interaction between an automated guided vehicle (AGV) and an operator (picker) could disrupt the work pace of the worker and induce mental stress. Most studies have focused on physical stress, and few have addressed dynamic changes and individual differences in mental stress caused by AGV–picker interactions. This study experimented with the order picking process that incorporates interaction with AGVs by providing a voice notification regarding the number of shelves waiting in front of the workstation (WS) and shelf transfer delay to address this gap. A dynamic Bayesian network (DBN) was used to predict mental stress levels assessed by the picker after processing each order. Further, this study constructed group-specific DBNs for groups with high and low scores on the Big Five factors strongly related to robot acceptance (extraversion, agreeableness, and openness) to address individual differences in stress. The main finding revealed that mental stress levels were higher when the number of shelves waiting in front of the WS was greater or when delays occurred in transporting shelves; further, they were higher for workers with lower levels of extraversion, agreeableness, and openness. The DBN achieved the highest accuracy (0.758) compared to those of three conventional machine learning methods. In addition, the overall accuracy of the DBN (0.768), which was divided into groups with high and low agreeableness scores, exceeded this value.
Construction is a critical sector in the economy, but it has several inefficiencies. Modular construction techniques appear to tackle this issue by building modules of the buildings in a factory-like environment and then shipping them to the construction site for assembly with other modules. However, this brings logistical issues, as the transportation of the very high-volume modules is very costly, as special transportation requires several mandatory extras, and these costs are hard to estimate, especially due to the lack of data. This study develops a deep learning model for transportation cost prediction in modular construction. Given the lack of data on module transportation, we use transfer learning techniques to use knowledge from a previous deep learning model that predicted the cost of raw material transportation. We tested the proposed methodology against three other strategies, and the proposed model using transfer learning achieved the best performance. However, there is room for improvement as the predictive error is still quite high. The novelty of our study lies in its novel application, which was never considered before. This study also contributes to practitioners, as it allows module designers/architects to create modules with cost-effective transportation.
The evolution of Human-Robot Collaboration (HRC) has moved beyond simple task-sharing, progressing toward advanced systems that integrate cognitive and physical dimensions in a proactive approach. The prevailing research is increasingly emphasizing adaptive interaction, predictive intelligence, multimodal perception and mutual synchronization between humans and robots. In this study, a literature mapping review was conducted to capture the latest trends and practices in HRC design and control, also distinguishing human-centric from robot-centric approaches. The analysis identified five thematic clusters capturing the current trends, i.e.: (i) AI-driven cognitive augmentation; (ii) seamless safety; (iii) enhanced digital twins, augmented reality, and virtualization; (iv) multimodal perception and mutual awareness; and (v) robot humanization. A literature key gap emerged, i.e., the lack of a holistic, system-centric framework that comprehensively integrates both human and robotic perspectives, in system design and control principles as per the identified dimensions. To address the gap, the Human-Robot Embodiment paradigm is introduced and proposed as the evolution of proactive HRC, structured around three integrated dimensions: (a) Situational Awareness and Continual Learning, (b) Enactment and Entrainment and (c) Bidirectional Control. The paper concludes by outlining challenges, opportunities and future directions for implementing the unified HRE paradigm.
The global challenge of balancing food supply and demand has intensified due to rapid population growth and climate change. Among various solutions, alternative proteins have emerged as a promising pathway to reduce environmental impact and meet rising protein demands. However, integrating these proteins into existing culinary systems presents significant design challenges. Current approaches often rely on simple ingredient substitution, overlooking the complex interactions among taste, texture, nutrition, and cultural acceptance. To address this gap, we propose a holistic systems architecting approach to create recipes that achieve new value propositions while ensuring sustainability. This requires a comprehensive redesign of food production and supply architectures based on Systems Engineering (SE) methodologies. Building on the framework of Digital Transformation (DX), we introduce the concept of Food Transformation (FX) to guide sustainable and innovative recipe design. The basic concept was initially introduced by the authors at INCOSE IS 2024. In this study, we further develop the FX concept by proposing a systems architecting approach focused on the explicit design of "deliciousness." The recipe design process is organized into three conceptual stages: Digitization-Foodiation, Digitalization-Foodilization, and Digital Transformation-FX. We demonstrate how each stage contributes to advancing sustainable and consumer-oriented recipe innovation through concrete examples.
The manufacturing industry faces an increasing demand for high product customization and high delivery performance in a volatile market environment. Companies rely on high-performing production logistics to meet these market demands, especially minimum TTPs. If extended TTPs are captured in production monitoring and control (PMC), root causes must be identified, and countermeasures must be initiated to avoid jeopardizing their market position. However, complex logistic cause-and-effect relationships often obscure the root causes. Logistic modeling enables a root cause analysis (RCA) based on generally valid cause-andeffect relationships. The increasing data availability unlocks the potential for identifying company-specific root causes using Machine Learning (ML). This paper presents an ML-based methodology for RCA of extended TTPs using clustering and regression algorithms. The methodology orchestrates the application of logistic modeling and ML in PMC to benefit from expert knowledge about generally valid and company-specific cause-and-effect relationships in RCA, thus improving the target-oriented derivation of measures. The methodology is validated with a tool manufacturer's production data.
This study addresses the limitations associated with the adoption of Operator 4.0, Operator 5.0, and human-centricity within manufacturing environments. Specifically, it examines the organizational barriers to the implementation of Operator 4.0, Operator 5.0, and the human-centricity paradigms. The study employed a mixed method, two-step Qualitative → Quantitative design where qualitative data analysis was integrated with advanced AI-based text-mining techniques. The findings indicate that the successful implementation of human-centricity necessitates addressing some fundamental organizational barriers, particularly those related to leadership approaches, operational efficiency, and the cognitive load of the operator. The results reveal that the most significant barriers are management-related, suggesting that technological solutions alone are insufficient without corresponding human-centric cultural changes. The categorization of organizational barriers into strong, moderate, and weak provides insight into the systemic and individual challenges that hinder the transition to a human-centric organization and the transition towards Operator 5.0 with a focus on the human-centric part.
As the manufacturing industry rapidly shifts toward high-mix, largescale production and extreme personalization, scheduling optimization has become significantly more complex than before. Existing search-based methods such as mathematical optimization, metaheuristics, and constraint programming face fundamental limitations in handling the high variability and uncertainty that characterize real-world manufacturing environments. Deep neural network technologies are emerging at the forefront. When coupled with the human-centric manufacturing paradigm of Industry 5.0, there is growing anticipation that Large Language Models (LLMs), capable of interpreting constraints and objectives expressed in natural language, could lead to more flexible and sustainable manufacturing ecosystems. In this study, we propose a method for solving manufacturing system scheduling optimization by utilizing an LLM, a form of unsupervised deep neural networks. Specifically, we focus on the Job Shop Scheduling Problem (JSSP), known to be an NP-hard task, and introduce a process wherein the JSSP is represented in natural language and then trained with an LLM using LowRank Adaptation (LoRA) and an enhanced prompting technique. Our results show that this approach achieves scheduling performance comparable to existing neural network methods, thereby suggesting the potential for LLM-based scheduling optimization to serve as a key tool in human-centric manufacturing under Industry 5.0.
Environmental challenges, such as climate change, energy inefficiency, emissions, and waste, pose significant threats to global sustainability. These issues are critical due to their impact on ecosystems, economies, and human health, necessitating innovative solutions. Despite extensive research, there is a gap in integrating system dynamics (SD) and machine learning/artificial intelligence (ML/AI) for holistic environmental management. This study addresses this gap by reviewing literature from 2014–2024 that combines SD’s feedback-driven modeling with ML/AI’s predictive and optimization capabilities. The findings reveal that this hybrid approach enhances long-term planning and decision-making in areas like waste, water, and energy management, though its application remains limited, highlighting opportunities for broader adoption.
The European battery industry is rapidly evolving due to demands for sustainability and digitalization. Large-scale battery production is essential for the energy transition but presents significant challenges, including in maintenance operations. By ensuring uptime and productivity, effective maintenance is key to industrialization. This study adopts a socio-technical lens to examine how sociological, technological, and organizational factors influence maintenance operations in battery production. Through ethnographic research within a real-world gigafactory, we gathered in-depth data on the socio-technical interactions of maintenance to identify critical challenges and establish important development needs. Thematic analysis resulted in the formulation of 31 distinct and relevant research avenues, providing industry and academia with strategic guidance and actionable blueprint for advancing maintenance operations in battery production.
This article aims to evaluate and analyze the relevant logistics aspects and propose the most efficient route to strengthen the competitiveness of the Brazilian soybean production chain in the international market. We found evidence that decision-making about the flow of soybean production is related to aspects of cost, time, losses, and emissions. This article uses the combination of AHP and ANP methods for decision-making regarding which logistics corridor will be used to flow soybean production. The decision model was created considering experts in the logistics area. Three different logistics corridors were considered: sending soybeans produced in Mato Grosso do Sul to the Port of Paranagua, the Port of Santos, or the Port of São Luiz. The results indicate that the Port of São Luiz is the best option for shipping soybeans produced in Mato Grosso do Sul. The criteria of time, cost, and losses in transportation were considered important indicators in the choice of the logistics corridor and, although the Port of São Luiz is the most distant from the soybean-producing region of Mato Grosso do Sul, it is possible to infer that its choice is due to the agility in unloading the soybean cargo, which reduces the idle time of the combinations and provides greater profitability. Finally, additional studies need to be carried out to discuss more efficient technologies that reduce emissions and minimize physical and financial losses in the shipping of grain production from Mato Grosso do Sul.
Energy efficiency in manufacturing processes is crucial for mitigating climate change and optimizing industrial operations. Therefore, managing energy efficiency in manufacturing by accurate power prediction plays a significant role in this effort. In particular, it is essential to effectively model the power demand of the milling process, one of the most common and energy-intensive manufacturing processes. However, the common power models for the milling process based on material removal rate (MRR) neglect the influence of tool wear, one of the most critical factors. Meanwhile, some power models that consider tool wear cannot be generalized across different materials and manufacturing conditions. To deal with these limitations, this study aims to propose a novel dynamic power model. Especially, we incorporate tool wear and workpiece material hardness as key factors in the power calculation. Then, for power simulation, we simulate the milling processes of three different workpiece materials (aluminum alloy, stainless steel, and titanium alloy) and apply the proposed power model to derive the time series of power demand. As the results of the power simulation, the milling process of titanium alloy, the hardest material, exhibits the highest average power and total energy consumption, whereas the peak power is highest in the aluminum alloy process, the softest material. These findings demonstrate that both tool wear progression and material hardness significantly influence power demand, impacting energy efficiency and tool replacement cycles. This study contributes to facilitating energy-efficient machining and informed decision-making in manufacturing operations by developing a more realistic and generalizable power model.
In recent years, research into decision-making problems using reinforcement learning in uncertain environments has been actively conducted. However, in the real world, when making decisions using reinforcement learning, not only is excellent performance required, but the basis for the decision is also required to be explained. The Markov decision process is a framework for expressing an environmental model for reinforcement learning. In this study, we apply dynamic programming to an inventory management problem formulated as a Markov decision process to derive an ordering policy. To explain the ordering policy, we use a local explanation method for the pair of state and action, and calculate the influence of each inventory amount. We generated an explanation of the ordering policy in natural language using a large language model. We also selected a kernel width that affects the results of the explanation method. The results showed that the explanation method can effectively present the influence of each inventory amount in the inventory management problem, and the results of the explanation method could be generated in natural language using a large language model. Additionally, the accuracy of the explanation method can be improved by selecting an appropriate kernel width. This study proposes a method to make it easier for people without specialized knowledge of inventory management to understand the ordering policy. It contributes to improving the explainability of decision-making problems using reinforcement learning.
The integration of artificial intelligence (AI) applications into the smart manufacturing environment of industrial companies is of increasing interest in research and practice. In addition to the technological changes brought by this integration, there is also a direct impact on shop-floor employees working with these AI-systems. To ensure successful human-AI collaboration, it is essential to identify the impacting factors on the employees. Drawing on a cross-industry case study with 40 expert interviews, this study provides insights into current practices of four industrial sectors: automotive, chemical engineering, mechanical engineering, and electrical engineering. Based on a combination of TechnologyOrganization-Environment (TOE) framework and Technology Acceptance Model (TAM), this study compares current applications and indicates implications for human-AI collaboration from an organizational and individual perspective. The results show that shop-floor employees in different industries are influenced by several key characteristics for human-AI collaboration, such as the underlying digital infrastructure and corporate guidelines. Subsequently, the findings are discussed in the context of the differences between the individual and organizational levels, managing expectations, and supporting workload reduction. Thus, this study makes a significant contribution to the ongoing discourse on human-AI collaboration including the shop-floor level perspective.
