Catálogo de publicaciones - libros

Blood banks produce and store blood products in order to fulfil the uncertain demand at hospitals. Platelet pools are the most expensive and most perishable blood product having a shelf life of only four to six days. Production volumes need to be chosen carefully in order to reduce outdating while keeping the occurrence of shortages low.

We investigate the structure of the optimal production policy by solving a down sized periodic Markov Decision Problem. The optimal production volumes appear to depend on the number of pools on stock and their ages. Simulation results for the optimal -policy suggest two rules: the 1 and 2 rule. Both rules perform quite well. The 2 rule performs nearly optimal even if one acknowledges the distinction of multiple and limited compatible blood groups and the uncertainty in the supply by donors.

Die Steuerung der Fertigung von Chips in der Halbleiterindustrie wird aus Vorgabewerten von so genannten Performance-Indikatoren abgeleitet. Kennzahlen der Liefertreue werden dabei als Leistungsgrößen und Plan-Vorgaben verwendet. Um die Wechselwirkungen der Plan-Vorgaben in der internen Supply Chan einer Halbleiterfertigung zu untersuchen, wurde ein zeitdiskretes Modell entwickelt. Die Ergebnisse zeigen, dass sich die einzelnen Leistungs-Vorgaben gegenseitig beeinflussen und somit nicht unabhängig festgelegt werden können.

Our paper is dealing with the Capacitated Stochastic Lot-Sizing Problem. In addition to the usual model assumptions as stochastic demand and manufacturing times, cost for setup, we also consider cost for waiting and lost demand. The goal is to find release and sequencing decisions with minimal expected cost per time unit. To solve the problem we use simulation optimisation, i. e., we combine a simulator with a Parallel Genetic Algorithm. Some numerical examples show the applicability of the proposed approach.

Im folgenden werden flexible Fließproduktionssysteme mit Montagestationen untersucht. An solchen Stationen werden Komponenten von mehreren Zulieferstationen zur Bildung eines neuen Werkstücks zusammengefügt. Das ist die sog. Synchronisationsbedingung. Der Materialfluß ist asynchron. Die Puffer sind beschränkt. Die Bearbeitungszeiten sind beliebig verteilt. Der folgende Beitrag beschreibt ein Verfahren zur Abschätzung der Produktionsrate eines solchen Systems. Hierfür wird ein Dekompositionsansatz verwendet. Die betrachteten 2-Stationen-Subsysteme werden als Warteschlangenmodelle abgebildet, für die die virtuellen Ankunfts- und Bearbeitungsraten sowie die zugehörigen Variationskoeffizienten zu ermitteln sind. Die Approximationsgüte des Verfahrens wird mit einem Simulationsexperiment untersucht.

Supplier selection decisions at the strategic level are focused on strategic items with both a high supply risk and a high profit impact. Therefore, strategic supplier selection decisions have to be long-term orientated considering (i) mutual commitments between the partners involved, (ii) fixed costs upon selection of a new supplier in the form of investment in training, and technology, as well as (iii) significant costs of switching from one supplier to another. Existing approaches of supplier selection neglect the interdependencies in time arising from investment costs of selecting a new supplier and costs of switching from one supplier to another. Moreover, it is assumed that the set of employed suppliers can be changed each period without cost. These shortcomings of current approaches motivates the research presented in this paper. A stochastic dynamic model for supplier selection based on hierarchical planning approaches will be presented. This model enables the evaluation of alternative dynamic supplier selection strategies.

A major problem in modelling and subsequent simulation of process-oriented systems (ProC/B models), is the functional correctness of the model. Therefore a model should be first analysed for its functional correctness before it is analysed by simulation. Petri nets are well suited for model based and state based functional analysis, but are often not adequate or not used for the specification of process models. We present in this paper a transformer for an automatic mapping from ProC/B models onto PNs. The resulting PN-models can be analysed with PN-algorithms and the results from the PN-analysis can be interpreted at the ProC/B level.

Currently, many railway operators are increasing the frequencies of their trains. By condensing the timetable, routing trains becomes increasingly difficult as the chosen routes not only have to meet safety restrictions, but also guarantee some stability if delays occur.

We address the problem of routing trains through railway stations for a given timetable and outline two algorithms. The first algorithm searches for a feasible solution for the train routing problem based on an independent set modeling that is solved using a fixed-point iteration method. The initial solution is then amended by applying the second algorithm in order to increase the time slot of a chosen route, i.e. the time interval during which a train may arrive and find its designated route open. This algorithm is based on a local search optimization scheme.

Results showed that the fixed-point iteration found feasible solutions within minutes even for difficult cases, i.e. tight timetables. Though more time-consuming, the second algorithm allowed the average time slot length to be doubled, thus implying that it is possible to find routings which are more delay-tolerant. This helps to decrease impacts of late trains.

We describe a flexible indirect search procedure which we have applied for solving a special pick-up and delivery vehicle routing problem with time windows. The heuristic is based on an encoding of a solution as a sequence/permutation of tasks, a cheapest insertion decoding procedure, and, a threshold-accepting like local search meta-heuristic.

In many rural areas, the public bus service is demand-oriented: By far the biggest group of customers are pupils who are transported to their schools within certain strict time limits. Usually, all schools start around the same time, which causes a morning peak in the number of deployed buses. However, schools are allowed to change their starting times within some interval. The question is, how to simultanenously rectify the starting times for all schools and bus trips in a certain county so that the number of scheduled buses is minimal. This problem can be formulated as a vehicle routing problem with coupled time windows (VRP-CTW), which is an extension of the vehicle routing problem with time windows (VRP-TW), where additional coupling constraints on the time windows are introduced. We give a mixed-integer programming formulation for VRP-CTW, and present solutions and lower bounds for randomly generated and real-world instances.

Disruption management for airline crew schedules is important for the airline industry, since an increasing amount of disruptions to the regular operations occur frequently. The emphasis of this task is put on quickly obtaining one or more reasonable, at best optimal, recovery solutions from current disruptions, which has to be achieved within an acceptable time period. In this work, we propose a decision support framework that combines exact optimization methods and meta-heuristics for solving real-life practical problems. An exact method, based on a Column Generation type of procedure, is studied and tested, while a dedicated Genetic Algorithm working with a local improvement procedure provides the capability to solve the problem alternatively. Notably, a so-called procedure is applied to customize solution methods.