Conference Agenda

The classical two-dimensional bin packing problem is to put small rectangular items into larger rectangular bins without overlapping, so that the items are completely inside a bin and a minimum number of bins are used. In this paper we want to study a variant where the items have a given demand, i.e. the bins must contain a certain number of copies of each item type. In addition, the number of layouts should also be minimized in order to reduce changeover times. A layout is an arrangement of items within a bin. The primary optimization goals are to minimize the number of bins used and the different layouts. To achieve these goals and to better utilize the occupied area of the bins, additional optional items can be used. However, the use of these optional items is associated with costs. We also have to consider distance constraints. Each item must be separated from all other items and from the bin boundaries by a given distance associated with the item. The problem was motivated by the company Precoplat/MicroCirtec GmbH based in Krefeld, Germany, in an effort to reduce changeover times and waste in multi-layer printed circuit board production.

The energy transition entails a shift from a centralized electrical generation system comprising a few large power plants to a distributed energy resource (DER) landscape comprising millions of grid-connected DERs across various grid levels. The distributed nature of this future energy system necessitates the development of operation and control algorithms that reflect this trend. Consequently, a growing research stream is emerging in the area of distributed optimization algorithms. In particular, when dealing with DER in medium- and low-voltage grids, the local grid infrastructure plays a crucial role in determining the physical limitations of the desired algorithms. Optimizations have been developed that inherit the grid constraints and prevent congestion. In this work, component-based dual decomposition is used to separate the nodes and lines in a grid, leaving separate optimization problems for the individual nodes and a problem set for the lines. This reflects the individual agents at the nodes and the grid operator. The optimization is implemented as a distributed time series optimization algorithm to account for battery electric storage systems at the nodes and is solved using consensus “Alternating Directions Method of Multipliers”. The algorithm is implemented in a laboratory-scale microgrid. The experimental setup is designed to orchestrate the available flexibility in the microgrid in order to provide grid services to the local or upstream grid. Possible improvements to the convergence behavior are analyzed. Furthermore, the results of the communication impact on the algorithm and management of reactive power are presented.

Since many years, automated guided vehicles (AGVs) are used in industry to realize transport tasks. Additionally, many methods for planning and control of AGVs have been proposed in literature. While most AGVs are designed to transport a single load, multi-load AGVs gained increasing interest in recent years. They are able to transport multiple loads simultaneously resulting in more complex planning and control compared to single-load AGVs. With regard to material handling of multiple loads, the arrangement of the loads on a multi-load AGV is of particular importance when it comes to operations. The way multiple loads can be arranged depends on the design of a multi-load AGV which might entail loading constraints. For example, loads may be stacked on top of each other, restricting material handling of loads that are located below others.

To evaluate the performance of different multi-load AGV designs, we first formalize and present different design options. Second, we introduce a modular optimization problem which handles the purpose of scheduling multi-load AGV operations and allows to consider different designs. To achieve results for the optimization problem, we third present a solution approach that enables to address the computational complexity of large problem instances. Based on results from numerical experiments, we then analyse the performance of different multi-load AGV designs. On the basis of this performance, we finally derive insights and recommendations regarding planning and control of multi-load AGVs with different designs.