In liner shipping ships follow a fixed route within a fixed time schedule. Additional costs are incurred when ships are delayed. Delay can be handled in two different ways: prevention and recovery. To prevent ships to incur delay, buffer times can be added to the schedule. The buffer time will then be used to capture (a part of) the obtained delay. However, when a ship does not encounter delay during the trip, it has to wait before it is allowed to start with the new trip. The total amount of buffer time that can be allocated to a schedule is limited. On the other hand, ships can perform recovery actions to reduce the amount of delay. Examples of recovery actions are increasing the sailing or port handling capacity. In the ship delay recovery problem the objective is to determine a recovery policy that minimizes the costs associated to delays and recovery actions for a given liner shipping route with variable buffer times.

Order acceptance decisions have a significant influence on the short- and long-term performance of a firm. Firms, therefore, should make complex order acceptance decisions with regard to their organizational goals in an effective way. Models are called for to provide insight for decision-makers in developing proper strategies for order acceptance. This research presents an agent-based model to support order acceptance decisions in a multi-plant enterprise, particularly incorporating buyer-seller negotiations. The negotiation process is conceptualized in three phases, taking into account salient viewpoints on making order acceptance decisions. The conceptualization and the developed model not only provide valuable insights into the nature of negotiation in a supply chain setting, but also contribute to the understanding of strategy development for order acceptance.

When using combinatorial models for real-life problems, we often have to cope with uncertainty in our data. Robust optimization tries to take this into account, by demanding that the solution to our problem is feasible with regard to a set of scenarios for our uncertain data. Recoverable robustness extends this idea by adding recovery: we may recover from an initial solution to some recovery solution by some restricted recovery method.
We present a column generation framework for recoverable robustness, introducing two decomposition methods that both return separate problems for each single scenario. We apply the framework to recoverable robust variants of some well known problems, like the knapsack problem and the shortest path problem. We show that we can find the optimal solution for these recoverable robust problems using branch-and-price and the classic algorithms for these problems.
Furthermore, extensive experimentation is performed on the size robust knapsack problem, comparing these column generation techniques to pure dynamic programming and local search algorithms. We show that the separate recovery decomposition yields some practical results in solving these problems in reasonable time and scales very well with the number of scenarios.

In this presentation we tackle portfolio selection problems which aim at selecting a subset of alternatives considering not only the performance of the alternatives evaluated on multiple criteria, but also the performance of portfolio as a whole, on which balance over alternatives on specific attributes is required by the Decision Makers (DMs). We propose a two-level method to handle such decision situation. First, at the individual level, the alternatives are evaluated by the sorting model ELECTRE TRI which assigns alternatives to predefined ordered categories by comparing alternatives to profiles separating the categories. The DMs' preferences on alternatives are expressed by some assignment examples they can provide, which reduces the DMs' cognitive efforts. Second, at the portfolio level, the DMs' preferences express requirements on the composition of portfolio and are modeled as constraints on category size. The method proceeds through the resolution of a Mixed Integer Program (MIP) and selects a satisfactory portfolio as close as possible to the DMs' preference.

When multiple items are auctioned sequentially, the ordering of auctions plays an important role in the total revenue collected by the auctioneer. This is true especially with budget constrained bidders and the presence of complementarities among items. In such sequential auction settings, it is difficult to develop efficient algorithms for finding an optimal sequence of items that optimizes the revenue of the auctioneer. However, when historical data are available, it is possible to learn a model in order to predict the outcome of a given sequence. In this work, we show how to construct such a model, and provide methods that finds a good sequence for a new set of items given the learned model. We develop an auction simulator and design several experiment settings to test the performance of the proposed methods.