This thesis investigates ways to optimise conveyor systems in intralogistics, focusing on improving overall efficiency and safety using simple as well as advanced modeling techniques. Recognizing the pivotal role of conveyor systems in material handling, and thus in the modern world, especially in sectors like parcel delivery and sortation, this thesis aims to address the challenges associated with systems design and operation. Developing an easy to use model that simulates sorter chute dynamics, and employing neural networks to reduce mis-sort incidents, this research provides practical solutions that can be, and have been, readily applied in industry. These efforts are directed towards enhancing the performance and safety of conveyor systems, thus contributing to the operational excellence and sustainability of intralogistics operations. Through this work, the author aims to blend theoretical insights and practical advancements, and emphasize the importance of model-based approaches in tackling the complexities of modern material handling systems. Material handling, as a foundational element of contemporary logistics operations, plays a pivotal role in dictating the flow, accuracy, and overarching functionality of various industries. Its significance is especially prominent in the post and parcel sector. This doctoral research navigates the realm of intralogistics, emphasizing the intricacies and opportunities presented by sortation systems, whether sort-centres, cross-docking facilities, or parcel hubs, that leverage crossbelt sorters. Despite the transformative potential of these sorters in revolutionizing parcel distribution, they are not devoid of operational challenges. Identifying and addressing these challenges constitutes the primary objective of this thesis. The research journey commences with an in-depth examination of current strategies shaping material handling. Through a rigorous assessment, the relevance, technical feasibility, and objectives of these strategies in the context of cross-docking operations are articulated. This extensive analysis not only illuminates the current landscape but also paves the way for subsequent investigations. Central to these explorations is the formulation of a mechatronic transfer model for a sorter chute. This model utilises the First Law of Thermodynamics, particularly highlighting the efficacy of brake belts equipped with 24V MDR drives. The nuanced relationship between design intricacies and the safety of operating personnel emerges as a recurring theme, underlining the equilibrium that must be struck between operational efficiency and safeguarding human assets. To ensure the real-world applicability of the proposed model, an empirical evaluation is undertaken. This phase involves juxtaposing the model's theoretical predictions against tangible data derived from industry operations. The findings from this juxtaposition underscore the model's value and its potential as a transformative tool for enhancing sorting operations. Changing the focus, the research then delves into the prevalent challenge of miss-sorts in cross-docking facilities. Innovations aimed at their minimization are introduced, laying the groundwork for a more streamlined, efficient, and ecologically responsible operational paradigm. Neural Networks emerge as a particularly promising avenue in this exploration. Their unparalleled capabilities in image recognition and classification make them apt for detecting instances where multiple parcels are loaded onto a single crossbelt carrier. Building upon this foundation, the thesis advances to design an innovative system that actively prevents double-loaded packages from navigating through the sorting process. The journey from conceptualizing this system to its real-world deployment and subsequent performance evaluation amalgamates the theoretical research with practical applications. The theoretical part of this research delves into areas of possible optimisations in various building types common to material handling, assuming mostly a hub-and-spoke network architecture. A discussion of various building types and their respective roles within the logistical network is followed by specific fields of conceivable improvements within those types, and a discussion of the weight of such hypothetical areas of improvement. This approach improves the yield of any engineered approach to improve processes, sorting early in the engineering process the most impactful areas and putting them into the focus of this research. Closing on the decision on where improvements have the highest impact, that decision being made also taking into consideration the impact on efficiency and putting this into context of customer satisfaction and ecological impact, this research develops an approach to use a Convoluted Neural Network to detect a defect in the sortation process and preventing this defect from having an impact outside of the facility, thus diminishing the impact of this defect on the environment, consumer satisfaction, and the operator of the facility. Presenting this approach developed within this research, this work also documents the real world implications of the result of this thesis when implemented in a real world environment, showcasing the real, measured impact this approach has when deployed in a logistic network. This doctoral thesis serves as a comprehensive discourse on the multifaceted domain of material handling within intralogistics. Merging methodical analyses, innovative modelling approaches, and empirical evaluations, it furnishes insights and actionable solutions poised to significantly influence the post and parcel industry. This research not only presents advancements in efficiency and safety but also underscores the imperatives of sustainability, setting the stage for a future where logistics operations harmonize with both human and environmental well-being.
| Date of Award | 13 Mar 2024 |
|---|
| Original language | English |
|---|
| Supervisor | Andrew Ball (Main Supervisor) & Karsten Schmidt (Co-Supervisor) |
|---|