Because the problem addressed here is a variation on the job shop scheduling problem, we can re-use much of modeling infrastructure for job shop problems. Rather than changing the static definition of a problem instance (that is, by predefining the resource alternatives and processing times for an activity), we use the original job shop instance as one resource alternative and dynamically generate the other alternative and processing time using a random number generator. Obviously, this is a somewhat artificial way to generate a problem, however it does demonstrate the primary difference in modeling from standard job shop models: the use of alternative resources sets and the specification of the differing processing times of an activity depending on the resource to which it is assigned. Readers who are uncomfortable with the artificiality of this example are invited to imagine replacing the calls to the random number generator with database accesses which retrieve the "real" data.