Constraints

A constraint is a rule the plan must respect. If it cannot be met, the result says the plan is not feasible, which rule fails and by how much (see Result of a plan).

Structural (always active)

They come from the data model itself:

Rule	Where it comes from
Precedences, with minimum and maximum waits	Tasks and processes
Capacity of each resource	Resources
Tasks in the same order do not overlap	Tasks and processes
Earliest start	Material or information that arrives on a given date
Fixed tasks	In progress, fixed by hand or known stoppages
Alternative routes: a single branch	Tasks and processes
Horizon	The plan does not go past the end date

Configurable

Same unit per order

All the tasks of an order are done in the same unit at one level of the hierarchy (for example, the same plant). Parameter: the level.

Interval between services

Between two services of the same type on an asset, at least the minimum and at most the maximum of its interval must pass. It can be counted in calendar time or in the asset’s available time (not counting the days it is already down).

Assets down at the same time

At most n assets under maintenance at the same time. Guarantees a minimum available to operate.

Capacity per service type

At most n services of the same type at once.

Capacity per group

At most n tasks at once in each group, for example in each workshop or each base. The group comes from a property of the task.

Fixed windows

Services that already have a date are respected.

In shifts

In absence cover the rules are labour rules: minimum rest between shifts, consecutive days, weekends off, skills per post, 12-hour shifts and reinforcement quotas. See Shifts and absence cover.