Lufthansa Technical Training GmbH (LTT) runs training courses for Lufthansa
Technik AG and for several other international airlines. It offers about 6
70 types of courses of which several hundred take place each year. The cour
se-scheduling problem LTT faces is to develop a schedule for the year that
maximizes its profit while meeting complex precedence, temporal, and resour
ce-related constraints. A good operational schedule should also meet a numb
er of subordinate objectives. In the past, LTT did all its scheduling manua
lly. We developed a prototype course-scheduling module, using the operation
al planning situation of 1996 as our major test instance. It turned out tha
t most of the schedules generated were substantially better than the soluti
on LTT constructed manually, both in terms of profit and computation time.
We subsequently developed a fully fledged decision-support system.