Crew Pairing

Crew pairings are created with the productivity targets as defined in the manpower plan keeping in mind the KPIs as specified by the organizational goals.

Crew pairings are the blueprint of the airline’s crew network. Pairing generation influences overall crew costs/staffing, operational reliability, and crew quality of life. Crew pairings must delicately balance the needs in these three areas. Prioritizing one at the detriment of the others may result in unintended consequences. For example, a tightly wound-up crew plan prioritizing minimization of staffing and pay hours may unravel in the daily operation and cause higher actual crew costs, disgruntled crew, and flight delays or cancellations.

Airlines may also use crew pairing generation in earlier parts of the process. For example, the airline may build a pairing solution to evaluate proposed flight schedules before publication, to give feedback to network planning on routings/flight timing, to evaluate proposed work rule changes, or to give inputs to the manpower planning process.

Pairing Generation Input

• Flight schedule in its entirety, including aircraft routings. Even if the entire flight schedule will not be covered in the pairing solution, the flights should be included for purposes of available deadheads.

Planned aircraft routings are important especially for short-haul networks. The airline will typically want the crew to follow the aircraft within a duty period to minimize crew-related delays or the need for re-crewing a flight in the daily operation. However, crew aircraft changes will often allow the airline to operate more efficiently, maximizing the efficiency of a crew’s planned duty period or pairing. A good collaboration between aircraft routing and pairing planning helps to gain maximum efficiency.

• Other airline (interline) flights or forms of transportation that can be used for positioning between working flights. Some airlines rely on interline deadheads for planned pairings, although these can introduce operational complexity and dependencies outside an airline’s control.

• Flights specified to be covered by the pairing solution– “what are you solving?”

• Date range–this is based on the airline’s rostering period. Most airlines solve pairings by calendar month. Solving for longer date ranges locks in the flight schedule. Depending on the airline’s crew work rules and technological capabilities, longer rostering periods make close-in flight schedule modifications costly or infeasible.

• Crew category for which you are solving (cabin crew, pilot, all or subset of either). Typically cabin crew and pilots will have different union work rules and/or governmental regulations on duty and rest. In Europe, governmental regulations on duty and rest are the same for cabin crew and pilots. However, this is not the case in the United States, and some other parts of the world.

An airline may wish to solve all crew categories together for operational efficiency and crew coordination. This requires using the most restrictive rules for either category. A large airline with restrictive union rules and multiple pilot crew categories may wish instead to solve pilots and cabin crew separately. This typically results in a more cost-effective planned solution but can cause operational complexity.

• Equipment type–some airlines solve all equipment together; some solve separately by crew qualification. For example, an airline with multiple fleet types will have a separate solution for each separate crew category.

• Flights already covered by previous rostering period’s solution–this is especially important to ensure no flights are over- or under-covered. This usually consists of flights in the current rostering period covered by pairings starting in the previous rostering period (if the airline is building multi-day pairings). Depending on the crew work rules, the planner may allow the pairing generation process to modify duty periods of pairings beginning in the previous rostering period (within certain guidelines).

• Any other special logic to specify what is to be solved in a particular solution. For example, the planner might create a subset of pairings for a specific qualification such as cabin crew language. Some airlines may solve domestic and international pairings separately.

Rules – “What is Allowed?”

Rules are binary and limit the possibility of duty periods and pairings the planner (or optimizer) is allowed to build.

• Duty and rest limitations, the more restrictive of the governmental regulations or the airline’s rules (often dictated by union agreements). This includes the maximum planned duty period and block time, and minimum planned rest, for the crew being solved, minus any buffers the airline chooses to use for operational insurance.

• Maximum pairing length (number of days)–this is typically driven by crew work rules and cost. For example, a union contract may specify maximum pairing length of four (4) days. An airline may elect to build only single duty period pairings if the flight schedule allows, to avoid hotel costs. Typically, longer pairings will result in higher hotel costs but fewer minimum pay costs and crew aircraft changes. Long-haul international airlines with fewer crew bases will need longer pairings than short-haul airlines with multiple crew bases.

• Crew aircraft change rules (under what circumstances crew is allowed to change aircraft on a given day, minimum and maximum times. Shorter aircraft changes result in more efficient planned pairings but will be more susceptible to crew-related flight delays. Long aircraft change time will create inefficient ground time in the planned pairings. Crew work rules may influence what values are used.

• Crew bases, including whether multiple airports are grouped together for a single crew base, and the logic of starting/ending associated. For example, if a large metropolitan area contains three major airports, the crew work rules may allow pairings to be built in and out of these multiple airports, sometimes with ground transportation added.

• Other business rules, including union agreement or operationally driven rules not listed in the above categories. Business logic that is not a “hard rule” should be used sparingly in rules. Sometimes a high-priority business logic parameter may be more appropriate. For example, rather than saying “no crew overnights allowed in XXX city”, the planner should instead use business logic driving the solution to only allow overnights in XXX city if it’s the only feasible way to cover the flights, given the other rules.

Business Logic – “What is a Good Solution?”

Logic guiding quality of solution within given schedule and rules.

• Crew quality of life considerations: are certain types of pairings preferred by crews? This could be pairing length, fatigue considerations, and other human factors items. For example, crews may prefer longer duty periods at the start of a pairing and shorter duty periods at the end of a pairing due to fatigue.

• Operational considerations: how much buffer to plan for rest, duty, and crew aircraft change time to minimize pairing breakage or flight delays. The airline might also have operational parameters to avoid other things that introduce operational risk. For example, they may want to heavily penalize crew aircraft changes in a particular city or a particular time of day. Advanced pairing generation technology may allow use of flight on time performance forecasts to determine optimal solutions.

• Cost considerations: includes pay calculation for each duty period and pairing and other solution-related costs. This includes the value of working flights (usually block hours), value of deadhead/ground transportation, RIGs (“rate in guarantee”–minimum pay for a duty period or pairing above the flight credit), crew per diem, and hotel/transportation costs. The planner may indicate specific values for each of these and how heavily to consider them relative to other categories.

• Staffing considerations: limitations on staffing, often by day, equipment type, base, and qualification–set from manpower plan. To set staffing parameters, the planner must understand the relationship between the manpower plan (how many crew are available in each category/base) and how the pairings fit into crew rostering. Since hiring and base transfers often occur during a specified rostering period, the pairing solution should “smooth out” the flying allocated to each category during the rostering period as much as possible.

• Biomathematical FRM modules. If available the airline can use one of the biomathematical modules available on the market to either ensure that during the pairing optimization pairings are constructed that are not too taxing on the fatigue levels of the crew, or that the outcome of the optimizer is validated for fatigue levels or alertness levels.

• Some airlines may also use templates or mapped-in pairings to “lock in” desired pairings for a particular solution. For example, if a certain pairing is desired each day its flights operate, a planner could use this during pairing generation as a firm requirement or suggested template.

Pairing Generation Process

Due to the complexity of creating crew pairings, most airlines use pairing optimization software for this process. The optimizer will take in the parameterized rules and business logic to produce one or more solutions. Depending on the size of a solution (number of aircraft, days included in solution, complexity of rules, number of possible combinations of flights), this could take minutes or days.

In larger pairing solutions, a planner may want to start by optimizing a given day or week of the rostering period, then map that into a larger solution. Most airline networks vary enough within a week or month to make this process less helpful. In most cases, an airline will want to optimize the whole solution at once.

Most pairing optimizers give planners the opportunity to assign values to each parameter. A planner may choose to run a variety of solutions with different parameter values. They may choose to use different rules for duty and rest by season to give different buffer levels. This is also how the planner drives the solution towards the airline’s goals. It allows the planner to prioritize certain outcomes over others. For example, stronger weighting might be put on staffing parameters if the airline is short-staffed.

A planner may run many different optimization scenarios with tweaks to optimization inputs before selecting a final solution for the rostering process. Rather than starting a new solution “from scratch”, a planner might send a solution back to the optimizer with some tweaked parameters and re-run it with that as a guide.

Once a solution has been selected, the planner will review to ensure flight coverage, rule compliance, and quality of solution. A planner may use separate systems for various parts of the review process.

A planner may choose to make manual changes to the pairings prior to sending them to the rostering process.

Pairing Generation Output

• Pairings: specified flights should be covered in pairings made up of at least one duty period each; pairings will be used later in the rostering process

• List of any uncovered flights: A valid pairing solution should cover all the specified flights, as that is the pairing optimizer’s main objective. However, some flights might be intentionally left uncovered to be handled in a later process, or due to solution infeasibility within specified rules. For example, if the maximum pairing length is set to one, but the optimizer cannot solve the flight schedule without planned crew overnights, the flights will be uncovered in the solution.

• Overnight/layover information for hotel and transportation planning. This could be as simple as the number of planned crew overnights in each city, or a detailed format required by the airline’s hotel procurement processes.

• List of deadheads, including interline deadhead and other transportation bookings needed. The airline will then need to book reservations on these flights for the crews. This may happen downline in the rostering process when crew names are assigned. Depending on the timelines and booking rules, the airline may wish to hold seats without names before the rostering process occurs.

• KPIs and reports based on business logic. This is usually a combination of crew cost, operational metrics, and quality of life. Standard KPIs are usually total pay hours, total pairings/duty periods (by base/category), average pay + block time per duty period, average pairing length, average duty day, and total crew aircraft changes. The pairing optimizer may provide multiple reports for more granular quality control.