DSL = Domicile Scheduling Letter (monthly scheduled IDs/pairings)
The complex process that ultimately produces a flight attendant DSL begins in the Aircraft Scheduling department --- where the interaction of a wide range of variables determines how United's fleet will be deployed for any given operational period. Among the factors affecting which planes will fly where, when, and how often are: customer demand, revenue potential, maintenance requirements, station facilities, slot and gate availability, competitors' tactics and pilot manpower allocations.
The computer program (actually a series of programs) used in DSL construction for both flight attendants and pilots is called ACRUZER (a loose acronym for Airline Crew Schedules Optimizer). Small, straightforward pairing problems (the domestic wide-body DSL for instance) can usually be solved manually. The 1,500 segments in a typical narrow-body DSL, on the other hand, could never be paired effectively without computer assistance. There are at least 28 trillion different ways to pair those 1,500 segments. The domestic ACRUZER optimizer is smart enough to reduce that staggering volume of choices to a manageable pool of feasible low-cost pairings covering every operating segment. Given a few hours, the optimizer will quickly move beyond simple coverage considerations to improve the efficiency and quality of its pairing solutions.
Although ACRUZER was designed to handle both domestic and international, it is not utilized for international ID creation as many variables (staffing, contractual rules, etc.) make it difficult for the program to develop an efficient solution. As a result, the international creation remains a manual process.
Usually, aircraft utilization and routing decisions are finalized about 60 days prior to a schedule's start date. The Crew Resource Analysis and Forecasting Team (CRAFT) then determines staffing levels for all flights. Once complete, the Advance Schedule Planning (ASP) crew scheduling analysts can begin to build the flight attendant DSL.
Three essential objectives drive the DSL construction process: cost control, operational feasibility and flight attendant working conditions (quality of life).
Costs - Onboard analysts track - and attempt to reduce - all excess costs associated with flight attendant pairings. Excess costs are defined as all DSL expenses over and above flight attendant salary. These costs include flight time credit, hotels, meal money (per diem), and transportation (limos, shuttles, vans, etc.). On average, these costs currently amount to about $8.5 million per month.
Data files within the ACRUZER program allow analysts to enter actual dollar costs for all these variables. The optimizer uses this economic data to find the lowest cost solution for any given pairing problem.
Operational Feasibility - The value of low cost pairings can be diluted or dissipated entirely if these cost efficient pairings can't be flown consistently in the real world. To ensure that IDs can actually be flown as built, several operational "cushions" are typically included in ACRUZER set-up files. Cushions most likely to improve operational performance include an extended 8-in-24 buffer (for domestic) and longer minimum layover and connection times. In addition to general system-wide cushions, connection, layover and hotel parameters are often tailored for specific stations, equipment, or time of day.
Flight Attendant Quality of Life - Although the Collective Bargaining Agreement clearly mandates specific limits for all working conditions, additional consideration of flight attendant quality of life issues must be integral elements of the DSL process. In other words, the reality that some perfectly legal pairings are extremely undesirable must be acknowledged and addressed.
Generally, problem pairings involve night-into-day issues, an inequitable distribution of attractive (or unattractive) flying, long sits, or long duty days preceding or following short layovers.
In cases where only a limited number of scheduling options exist, there may be no reasonably priced alternative to a pairing with obviously difficult working conditions. In the large narrow-body DSL, however, there may be several efficient pairing solutions with total costs within a few hundred dollars of one another. The flexibility afforded by this large pairing pool sometimes allows analysts to improve working conditions at little or no cost.
The Crew Resource Analysis and Forecasting Team (CRAFT) determines optimal staffing levels for all flights. Staffing programs pull information from several sources:
Staffing forecast programs, one for domestic and one for international, combine all of the above data into detailed reports incorporating all the information needed to construct trip pairings. The international report is organized by region, and the domestic report is organized by fleet type.
Since base crew complements rarely change for North America flights, the process is fully automatic as far as determining the number of flight attendants to schedule on each flight. Base domestic crew comps are 6 or 3. One-position supplements are often necessary to ensure that each aircraft is staffed to FAA or contractual minimums. That means that there are least two 1-position IDs to match the base of 6 for the 777s, and at least three 1-position IDs for the 747s. Also included in the 1's file are any pre-scheduled XSCs needed as a result of passenger load forecasts.
The international schedule requires variable staffing. That is to say, the size of the base crew complement can vary from month to month and from domicile to domicile. For this reason, the international staffing and ID building process is manually intensive. The international schedule is not as large or complex as domestic and, for the most part, it is flown the same way from month to month. CRAFT reviews the forecasted loads on both outbound and return international flights to determine optimal round-trip staffing. When international pairings do change, CRAFT revises staffing requirements to ensure that any XSCs required are added on the correct days.
Normally, Advance Schedule Planning (ASP) crew scheduling analysts begin to build a flight attendant DSL about 60 days prior to the start of the crew month.
Daily Pairings - To start the process, ASP analysts download aircraft segment and routing detail from master files updated and exported daily by Aircraft Scheduling. Next, planned staffing levels are obtained from the CRAFT group. This is also when current hotel and transportation costs are verified in the ACRUZER set-up files.
International analysts take the above information and manually build pairings using the previous month's DSL as a basis (since the international schedule does not change drastically from month to month). All crew complements, weekend and dated exceptions are included in this process. It is broken down into 3 distinct groups: Honolulu/West Coast (and Denver); Language; and all other international flying.
The domestic process begins with the creation of the daily pairings - a set of IDs covering all segments operating at least 4 days a week. The new month's solution is usually based on the previous month's DSL. Analysts apply the new aircraft schedule to the old pairings in an effort to maintain some continuity and to provide the ACRUZER optimizer with a "head-start" in its search for a good initial pairing mix. The rough draft DSL that soon emerges from this procedure is used to identify potential operational risks and extremely undesirable working conditions. Analysts then seek low or no cost solutions to these problem IDs. When fixes are found, the revised pairings are locked and stored. Several more optimizer runs then occur with an analyst review following each run. This strategy usually produces a reasonably priced, operationally feasible solution fairly quickly (2-4 days). Multiple optimizer runs thereafter incorporate late aircraft schedule changes and, eventually, a nearly optimal set of pairings is found.
It's a reiterative process. Each day, domestic analysts apply an updated version of the aircraft schedule to the previous day's pairing solutions. Internationally, any changes to the aircraft schedule are manually input into the files. IDs that are no longer legal or operationally viable because of aircraft changes are repaired, and the optimization/review cycle begins again. The cycle continues until about 40 days prior to the start of the crew month, which is when the balancing phase begins.
Balancing - In the initial stages of the DSL process, all ID building efforts have been occurring in a "best flown" environment. Basically, in the best flown mode, the ACRUZER domestic optimizer and the international analysts put flying where cost data indicates the flying belongs - with no regard at all for flight attendant availability at each domicile.
The balancing phase of the DSL process allocates flying to match active and available population levels at each flight attendant base. While adjustments in utilization can resolve some minor balancing problems, significant shortages or overages have only two remedies: flying must be moved or flight attendants must be transferred.
The CRAFT group decides which option is preferable based on historical data and future flying forecasts. If historical trends indicate that the "best flown" DSL consistently requires 60 more lineholders at Domicile "A" and projected flying levels for the foreseeable future seem unlikely to change dramatically, transfers to Domicile "A" may be offered. When no discernible trends are apparent, or if flying imbalances are seen as minor or temporary, flying will be moved to fit lineholder availability.
AFA Review - After the DSL is balanced, all domestic daily pairings and the complete international solutions are reviewed with the AFA's Central Scheduling Committee (CSC). The CSC recommends ID changes to alleviate difficult working conditions and to distribute attractive (or unattractive) flying equitably among domiciles. Typically, CSC recommendations involve about 5% of the daily domestic DSL. On average, analysts are able to improve 65-75 percent of these problem pairings with no significant impact on overall DSL costs. At this point, crew complements for the international DSL are determined.
Weekly Exceptions - Barring a major disruption of the aircraft schedule, the CSC review concludes the daily DSL process. The weekly domestic exception phase then begins. First, ACRUZER incorporates the actual operating frequency of all flight segments into the (theoretical) daily solutions. Pairings that really do operate with identical segments 4 to 7 days a week are identified and stored. Remaining segments are then covered in weekly exception pairings (Saturday only, etc.). Because weekly exception operations usually include a significant reduction in flying, deadheading is introduced at this stage of the DSL process. (While deadheading is costly, ACRUZER will identify circumstances where the alternative is even more expensive.) Unlike the schedule available in the daily DSL, the weekly exception segment pool is not rich in coverage options. Some pairings generated in this phase of the process aren't very efficient. It often becomes necessary to combine weekly exception segments with pieces of selected daily IDs to reduce overall DSL costs to an acceptable level.
The Final DSL - Upon completion of the domestic weekly exception pairings, the final phase of the domestic DSL process begins. Analysts apply single-dated flight cancellations and additions, transition divergent pieces of the monthly operation, and repair changes generated to account for charter operations, holiday reductions and mid-month equipment substitutions. Most daily and weekly IDs are excluded from this final processing, which is why single-dated transition, holiday or other exception flying is often expensive, or difficult to place in lines, or both.
Meanwhile, the international analysts break down the pairings into the pre-determined crew complements.
When the monthly exceptions are finished, the domestic and international DSLs are sorted, numbered and transmitted to the ASP group, the AFA Local Scheduling Committees, and to the support groups that generate hotel and deadhead requirements.
It is sometimes necessary to revise the final DSL to incorporate very late aircraft schedule changes or to accommodate line building needs. After these last minute adjustments have been made, the DSL is printed, and the pairings are loaded into the Crew Management System (CMS), where they will become "live" five days before departure. Absent a radical revision to the underlying aircraft schedule, the DSL process is now complete … unless, there is an end-of-month transition.
End-of-Month Transition - Several years ago, Onboard Service and Flight Operations convinced Aircraft Scheduling to begin new aircraft schedules no earlier than the fourth day of the crew month. This strategy eliminated almost all end-of-month (EOM) DSL transitions. (In other words, IDs from the old month can finish out their scheduled flying into the new month.) About twice a year, however, a new aircraft schedule will begin on the first day of the crew month. When this circumstance occurs, an end-of-month DSL becomes necessary.
The international end-of-month is done manually while the domestic crew schedule analysts use an interactive semi-automated approach. First, all EOM IDs that can operate unchanged or almost unchanged are identified and locked. Next, all other pairings that fly through the EOM are temporarily terminated at midnight of the transition date. ACRUZER programs assist analysts in rebuilding these disrupted IDs. Well over 90% of the single-dated EOM pairings created in this process are the same length (in days) as the flight attendant's original pairing - which makes them suitable replacements for the flying lost in the transition. When it is impossible to replace a lost ID with something similar, the value of the original ID is guaranteed with affected flight attendants becoming eligible for reassignment.
Q. Why are there so many one-day IDs?
A. In a cost-perfect world, all crew pairings would be one-day turns with at least 5 hours of flying time. These IDs are the ultimate in efficiency; no credit time, no hotel or transportation costs, and very low per diem expenses. Left to its own devices, the ACRUZER optimizer will build as many one-day trips as it can.
Q. Why does the DSL change so much from month to month?
A. In the DSL process, each month's crew schedule is a brand new puzzle. Although the process begins with an attempt to retain the previous month's IDs, a drastic remix of the puzzle's 1,500 pieces occurs more often than not. Usually, aircraft routing revisions, costs and domicile balancing requirements drive month to month DSL changes. Even when an attractive September pairing at Domicile A is 100% efficient, its components may be needed to solve 2 or 3 or 6 October problems in Domiciles B, C, and D. ACRUZER logic focuses on optimization of the entire DSL with little concern for the individual pairings that make up the optimal solution.
Q. Why is Domicile's A flying so much nicer than Domicile B's?
A. Frankly, there is usually a measure of luck involved. No domicile is ever singled out for reward or punishment in the DSL process. Cost, aircraft routings and flight attendant availability determine what flying will look like at each domicile. With up to 28 trillion pairing combinations possible, the ID building process is objective, but extremely random - not unlike the lottery.
Q. Why can't flight attendants and pilots be co-paired?
A. They can be, but it's very expensive. Even the limited co-pairing used in the United Shuttle operation often cost more than $100,000 per month. While some contractual differences between the two groups are troublesome in the co-pairing process, the real problem is that pilots are restricted to one equipment while flight attendants can work on any aircraft. This wide scheduling flexibility creates a rich pool of efficient pairing options. Millions of those low-cost possibilities disappear in a single-equipment DSL.
Q. Why don't crews stay with planes more often?
A. ACRUZER procedures do include incentives to keep crews with aircraft except when it's really economically advantageous not to do so. Currently, flight attendants stay with the plane approximately 30% of the time. Programming efforts are underway to increase this ratio.
Q. Are duty days longer now?
A. The average duty day is indeed longer - by 5% - since May 2003 when the new Collective Bargaining Agreement was implemented. Cost savings associated with the legality changes that allow longer duty days are part of the productivity increases granted by that Agreement.
Q. Why does an all-nighter continue on to another destination?
A. (Example: LAX-ORD and continue on to DAY) Duty periods that include an early morning segment after an all nighter should be rare. Usually, trips like this will be found only in exception pairings or in fleet-restricted DSLs where no other economical option exists. In the daily and weekly DSLs, analysts attempt to ensure that 'all nighters' are the final segment in a duty period.
Q. There's a nice DEN two-day trip wedged in the middle of a EWR three-day. It looks like both the DEN and the EWR pieces could stand alone. Why are they combined?
A. The pairing is a product of the balancing process. When a domicile's available population exceeds the flying assigned in the "best flown" pairing solution, it sometimes becomes necessary to "stuff" flying where it doesn't really belong. Conversely, when a domicile's population isn't sufficient to cover the flying that the domicile should "naturally" have, that flying must be moved elsewhere. In the example cited, it appears that the best-flown pairing solution gave DEN too much flying and EWR not enough.