The scheduler considers a plan infeasible when some constraint can't be met when attempting to schedule the plan. Tables 1 and 2 describe the various constraints used when attempting to schedule a plan. The demand constraints are associated with the plans that will be scheduled. The resource constraints describe the limitations of the common resources. Both sets of constraints are important to determining the feasibility of a plan.
Table 1: Constraints associated with Demands.
A plan enters as a set of demands for moving cargo (people are a type of cargo) from one location to another. The movements in the demands can have the constraints in Table 1 associated with them. The movements cannot begin before the ready date, and must be completed by the due date. Demands are satisfied by activities that take time at some to be determined interval. The activities that satisfy the demands can be ordered by defining a temporal relation. Two activities can be specified as one executing before the other, having the same start time, or having the same end time. Finally, given the type of the cargo in a demand, the demand can be split up into multiple sub-demands depending upon the partitioning constraints. The partitioning constraints available are cargo specific.
Scheduling a plan means assigning resources to the various activities that the plan describes. The description of these resources can include the constraints from Table 2 on how the activities that implement the user demands can be utilize the resources. The constraints in Table 2 are at differing abstraction levels of the resources. A ship might be available, but the capacity of the port where the ship is to be loaded might be exceeded. In this case, the scheduler is representing the port abstractly as an aggregrate capacity resource that allows the port resource to service multiple requests without temporal synchronization relying instead on some limiting capacity constraints. The different abstraction levels make scheduling a set of plans simpler.
Table 2: Constraints associated with Resources.
Producing a feasible schedule, that satisfies the physical constraints of the domain as represented by the resource constraints, is important. Thus resource constraints are held more important to preserve when scheduling than demand constraints. Thus all scheduling failures are failures of the demand constraints to live within the constrained resources. As an example, if all the airports are busy, then entering a plan to transfer some cargo by air will indicate a due-date constraint failure, rather than an airport resource-capacity failure. Mapping scheduling failures into reasonable actions thus requires information about the constraining resources.