We consider scheduling tasks at the United States Military Academy inWest Point. The 13 academic departments of USMA offer courses for the upcoming academic year. The USMA guarantees that all cadets are able to complete their academic program in 4 years. Hence the timetablingof the courses must provide a schedule for each cadet that contains all requested courses without a conflict. Furthermore, the academic activities of the student must be coordinated with military and physical activities. In addition to the term course scheduling we present the problem of term end exam scheduling where large set of exams taken by the students together with a small number of time periods will in general cause some conflicts. In order to avoid conflicts one could schedule, in addition to a course's primary exam, a second and if necessary a third makeup exam. A feasible exam schedule has to provide a conflict free exam period for all students.
The objective is to find an exam schedule with a minimum number of makeup exams.
These type of theoretical scheduling problems becomes much moredifficult and interesting if other 'real world' constraints and rules apply. In addition, for real world data there is usually no feasible solution. Hence in order to solve these "unsolvable" problems certain constraints have to be relaxed to overcome infeasibilties. We will present approaches to the scheduling problems which are based on a sequence of mathematical optimization models. This approach implemented in the Academy's Management System together with a substantial amount of human expertise was successfully applied for the last two terms at USMA.