Institute of Mathematics > Departments > Stochastic models in engineering science > Technology transfer > Deutsche Lufthansa AG

In **flight plan creation** the optimal frequency number for a destination with the optimal aircraft size and the best time slots are combined in order to optimize earnings. Since delays carry heavy costs, thereby departing from the economically optimal distribution of aircraft and personnel capacities, on the one hand provide the planned block and ground times with certain punctuality buffers and other, so-called operational reserves are kept for faults on the plan execution. The decisive factor is the volume of additional capacity. Because with the same buffer and reserve extent, the operational stability of two differnt flight plans can greatly differ.

Events that result in irregularities in the schedule implementing (for example, reducement of take-offs and landings at an airport due to bad weather) require operational interventions by the transport manager at the operations control center, in order to minimize the negative impact of regularity, punctuality and efficiency. These interventions are taken based on experience and intuition.

In order to evaluate the influence of parameters such as block and ground times and use of reserves and to compare different options for action for the transport manager, a simulator is developed which can be used both in the flight plan creation and implementation. Here historical block and ground times were evaluated and summarized in the corresponding clusters for important factors (for example, block hours after departure and destination airports and aircraft, ground times by airport and aircraft type). Then different schedules are simulated assuming certain conditions, to gain insight on the key performance parameters on the basis of this data.

The concept of **revenue management** is used in the aviation sector since the 1970s for integrated price and capacity control. In the meantime there it turned out to be a key to competitiveness. The goal is to find at any time in advance booking period the optimal trade-off between free-held seats for full paying late bookings and occupied seats for less paying early bird. Thereby one trys to find the highest possible generated result per seat kilometer.

To solve this problem there are various optimization approaches, which are to be compared in a created model world. These methods require a fixed capacity of seats. The existing methods are to be investigated for their suitability to the integration of a variable capacity, and possibly to develop a "new" optimization model that takes this into account explicitly.

Of course there are always passengers who do not appear for the flight, whereby more bookings are possible as a seat-capacity exists. Through such overbooking, it is accepted that passengers can not be transported (Denied boarding). Thus there is not transportation guarantee but a transportation probability which can be connected to a distribution assumption with the passenger seat load factor.

There is a dilemma between high utilization and a high transportation probability. The airline now bears the full utilization risk for each flight. There are considerations on how one can transfer the utilization risk to the passengers. Common risk hedges are considered from the financial markets and sought parallels to air traffic.

In the network planning of a regular airline there is the complex task of having to optimize not only individual flights, but all flights. To demonstrate the impact of individual change to the entire network, you need IT support. The IT tool used at Lufthansa is based on a logit model which enables attractiveness rating of flight connections. Various properties of the compounds are included in the review.