For the air cargo revenue management problem, this thesis studies a dynamic capacity control model on a single flight leg that is formulated by means of a Markov decision process. Applying revenue management techniques to the control of air cargo space involves various complexities. One important complexity is the uncertainty of capacity supply that arises as airlines transport both passengers and freight, with passengers having a higher boarding priority. A deterministic control policy simplifies this problem by assuming a full passenger utilization. Numerical experiments show that supposing the capacity to be random may improve the revenue compared to the deterministic approach. The experiments reveal that increasing levels of the expected capacity supply significantly affect the degree of revenue improvement, whereas the variance produces comparatively small deviations. The results are evaluated to be independent from the shape of the probability distribution.
