Agnieszka Sivertsen forsvarer sin ph.d.-afhandling
Kom og h?r Agnieszka Sivertsen forsvare sin ph.d.-afhandling 'Stowage Algorithms and AI for Energy Efficient Liner Shipping'.
Alle er velkomne.
Du kan ogs? f?lge forsvaret online p? Zoom >
Efter forsvaret er Institut for Mennesker og Teknologi v?rt for en lille reception.
Vejledere og bed?mmelse
Bed?mmelsesudvalg:
- Henning Christiansen, Professor, Institut for Mennesker og Teknologi, 真人线上娱乐 Universitet, Danmark (forperson)
- Catherine Cleophas, Professor in service analytics, Institut für Betriebswirtschaftslehre, Christian-Albrecht Universit?t zu Kiel, Germany
- Julia Pahl, Lektor, Institut for Teknologi og Innovation, Syddansk Universitet, Danmark
Vejledere:
- Line Reinhardt, Associate Professor, Institut for Mennesker og Teknologi, 真人线上娱乐 Universitet, Danmark
- Rune M?ller Jensen, Sealytix
Resumé
Liner shipping plays a vital role in the global economy. The e iciency, cost, and environmental impact of liner shipping are heavily influenced by the quality of the stowage planning being employed. The formalizes such planning in a combinatorial challenge that deserves more attention than it has hitherto received.
This thesis aims to infuse the academic research in container stowage with knowledge gained from years of experience developing real-life stowage planning solutions. The hope is to significantly further the state-of-the-art in academic container stowage planning while bridging the gap to industry applicability.
This thesis presents four main contributions: (1) A comprehensive literature review of the , (2) A representative problem formulation for the , accompanied by an extensive, real-life benchmark suite,(3) A novel decomposition search heuristic for the , (4) An e icient mathematical model for Template Planning that incorporates paired block stowage constraints.
Firstly, the literature review summarizes the state of container stowage research and highlights the lack of representative problem formulations and real-life data necessary for effectively bench-marking proposed solution methods.
Next, to address these problems, a novel problem formulation for the is introduced, designed to be easily scalable to industrial settings. This formulation includes essential but often-overlooked constraints, such as lashing and paired block stowage, which is demonstrated to have a strong impact on real-world stowage planning. The is presented with the most extensive benchmark suite known based on real-life data. Hopefully, this work will inspire further research in this area.
The third contribution introduces a novel decomposition search heuristic aimed at addressing the . This heuristic employs a decomposition method that significantly improves the quality of the obtained solutions. An in-depth analysis of the search heuristic with many solution modification strategies is presented. Furthermore, a machine learning-based feature selection method demonstrates how to quickly achieve vessel capacity estimates.
Finally, the thesis presents an e icient integer programming model for Template Planning, a subproblem of the , which incorporates the paired block stowage constraint. This subproblem has also been proven to be NP-hard by reduction from set partitioning.
Additionally, two mathematical models are introduced for bulk shipping. These models decide simultaneously on the routes, vessel assignments, and the amount of cargo to ship, all of which aim to minimize freight rates.