A comparitive evaluation of modeling approaches to the labor shift scheduling problem

The labor shift scheduling problem has traditionally been formulated using the set covering approach proposed by Dantzig (1954). The size of the resulting integer model with this approach, however, has been found to be very large to solve optimally in most practical applications. This paper relaxes the assumbptions of Bechtold and Jacobs (1990). This approach also requires a substantially smaller number of variables and non-zeros in the A-matrix than the equivalent set covering formulation.