ANSP roster prototype: the demand and leave

With staff and qualification records in place, the team defined shift types, derived staffing demand from the Sector Opening Timetable and distributed leave across the modelled year.

These were inputs for a simulation, agreed with the ANSP and its representatives. They are not a recommended pattern or set of limits for every operation.

Part 1 · Part 2 · Part 3 · Part 4

Define the shift pattern

The model used morning, afternoon and night duties. Each shift had a name, roster code, start and end time, minimum rest setting and relevant replacement allowance.

Two kinds of non-working period were distinguished. A 24-hour rest period after night duty could not be used for replacement work. The later off periods could be configured to permit replacement duties.

The resulting pattern was MANROO: Morning–Afternoon–Night–Rest–Off–Off. The rest period after the night was protected from duty; the two off periods were available for the agreed on-call arrangement.

Derive staffing requirements

The source SOT covered 24 hours per day over 104 weeks. The team reviewed minimum, maximum and average sector openings, including winter and summer demand.

The agreed assumption was three ATCOs per open sector plus two supervisors. Three open sectors therefore required 11 employees on duty in this model.

Two team-based staffing-requirement sets were prepared for the initial simulations: winter and summer, both using MANROO. Demand was lowest at night and higher during morning and afternoon duties.

This process translated a supplied sector-opening plan into staffing demand. It was not a machine-learning traffic forecast.

Place leave across the year

The simulation included annual leave, unplanned leave and English training. The starting allowance was 25 days, with an additional day for each year of service up to a maximum of 36 days. Carry-overs and additional allowances were also represented.

The average allowance was approximately 30 days per controller. Planned leave was arranged in blocks of 15 or seven to eight consecutive days, then distributed across the year for the simulation.

These leave assumptions affected the staff available in each month and therefore the resulting coverage and overtime.

Part 4 presents the first winter and summer results.