Correspondence: Anika Schroder, Institute of Mathematics, Clausthal University of Technology, Erzstrasse 1, 38678, Clausthal-Zellerfeld, Germany. Tel: +49 5323 722417; Fax: +49 5323 722304; E-mail: anika.schroeder@tu-clausthal.de

¹Michael Frank is senior project manager of revenue management development at Lufthansa German Airline (LH). Dr Frank's research interests include queuing theory, production planning and stochastic simulations.

²Martin Friedemann develops validation methods for hybrid forecasting as part of his work as project manager in the IT department at LH. He is also in the course of finishing his doctoral thesis, which will include an investigation of dynamic capacity allocation in revenue management.

³Anika Schröder is currently concentrating on the conclusion of her dissertation. She is sponsored by the research cooperation of LH and Clausthal University of Technology. Her research pertains to dynamic pricing and customer choice, including stochastic simulations for comparing different demand models.

Received 2 August 2007; Revised 2 August 2007.

Abstract

Simulations are commonly applied in Revenue Management (RM) to give proof of new concepts. Most publications in this area focus on the methodical part and disregard a detailed description of the simulation in use. The underlying assumptions and simplifications, however, have a strong impact on the results. Although the make-up of a simulation environment always depends on the investigation undertaken, some general rules for simulations in RM can be established. In this paper, we provide a guideline for developing a stochastic simulation model to be used for analyses in RM. Many occurring questions cannot be universally answered. Therefore, we point out difficult decisions and give advices concerning setup, input data and calibration as well as the modelling of demand, forecast and optimisation. Finally, we present a specific simulation study as an example.