Rostering

Edward Sitarski, Vice President of Technology (e.sitarski@schedule-masters.org)

Introduction

Rostering is one of the final steps of Public Transit Scheduling. After vehicle assignment (blocking) and crew assignment (runcutting), rostering is the process of combining daily runs to create complete weeks of work for drivers.

In North America, rostering has often been left to the drivers, where each driver picks his or her runs and days off in order of seniority. This practice often leads to "leftover” runs that cannot be legally combined and require costly part-time or overtime work. It can also lead to an excessive number of runs with non-consecutive days off.

In Europe, rostering is much more complex where the goal is to minimize the difference between individual driver’s workweeks.

For these reasons, is can be highly beneficial the transit scheduler to create all the rosters at once to minimize leftovers, and then let the drivers choose a roster from this set. In addition to minimizing leftovers, powerful software now exists that allows the scheduler a great degree of control over the type of rosters created. The scheduler can easily choose to maximize driver convenience or minimize the difference in work for all rosters or only for rosters of a certain type.

This paper outlines these cost-effective rostering techniques and how The Master Scheduler software can help achieve those goals.

 

The Goals of Rostering

There are a number of basic goals that are always considered at each property, some are required, and some are optional. These goals have different degrees of importance at a given property.

1. Required:Respect required days off in a week. If drivers work a 5-day week, then they need 2 days off per week. If drivers work a 4-day week, then they need 3 days off per week.
2. Required:Respect minimum off time between consecutive working days. Drivers need a minimum amount of time off when one day ends before work on the next day begins.
3. Optional:Maximize driver convenience. This includes maximizing consecutive days off, minimizing the difference in start time from day to day and minimizing combinations of straight and split runs (at some properties, this is not allowed at all).
4. Optional:Maximize driver stability. Here the goal is to try to minimize the number of different routes a driver has to work in the week. The idea is that this will result in higher customer service as the drivers do each run more often and learn how to drive it better.
5. Optional:Minimize part-time and leftover work. At some properties, part-time work weeks must be paid for at a full-time rate and/or leftover work must be paid at an overtime rate.
6. Optional:Minimize the weekly work variance between drivers. Especially in Europe, some properties have to pay a premium for drivers who work over and under the average week. In North America, it is sometimes desirable to make certain classes of rosters "roughly” equal in duration (for example, make all rosters with non-consecutive days off roughly the same amount of work).

Rostering Strategies

As is usual with problems in public transit scheduling, the optional rostering goals conflict, and it is usually impossible find a solution that satisfies all simultaneously.

Another issue is to determine whether driver convenience should be considered in order of seniority (for example, the most senior drivers get first pick at the most convenient rosters). Alternatively, it could be considered across the board for all drivers (for example, all drivers could start the next day’s work within one hour of the previous day’s start time). Practically, both strategies could apply in different situations at the same property.

The first step in finding a good roster is to decide which are the most important goals, and which are the goals that can be sacrificed for the important ones. There are a number of techniques that can be used:

Let the drivers pick their runs and days off

This technique is common in North America. Drivers meet with the scheduler - in order of decreasing seniority - and pick desired days off as well as the runs that they want to work. The remaining runs are left for drivers with lower seniority and the process repeats until all runs have been chosen. This allows the most senior drivers to get the best work weeks, and the first work weeks can be stable. However, it can lead to problems:

1. Many incompatible work days left over at the end. A number of work days cannot be combined together because they would violate the restriction of minimum time off. This does not mean that there is no solution that incorporates those work days, it just means what the drivers picked dud not lead to that solution.
2. Many "ugly” rosters at the end for low seniority drivers. For example, a high variance of start time between work days or a lot of mixing between run types in the same roster. This can, in some properties, contribute to driver dissatisfaction and higher absenteeism.
3. A high degree of inequity in the number of hours worked per week between drivers. Some may be too short to pass as "regular”, and others may be too long to be worked without overtime. This can also lead to driver dissatisfaction if the inequities are too great.

Leftovers can require part-time workers, or in some cases, the leftovers must be worked by full-time employees or as overtime. This can be expensive, and often there are union rules governing the number of part-time staff that can be used (sometimes none are allowed at all).

Generally this process takes a few weeks as each driver will have to choose his or her work week. Problem rosters are only discovered at the end of the process when there is no opportunity to make any improvements.

The Master Scheduler directly supports this process.

Let the drivers pick from a set of scheduled work-weeks

This requires the scheduler to create a set of feasible and complete rosters first, and then allows the drivers choose those given rosters in order of seniority. This allows the scheduler to plan the roster cleverly so that the problems of letting the drivers choose their own days off can be avoided. Of course, this means that some of the driver convenience be sacrificed, as the rosters will be designed without each driver’s direct input on desired days off and runs.

In the past, many properties did not use this strategy because there was no way to create a good set of rosters quickly. All this changes with computerized scheduling and The Master Scheduler.

There are a number of ways that properties may want to create rosters to maximize consecutive days off and minimize any leftover work:

Maximize driver convenience

Here the goal is to minimize the variance in daily work start-time for each individual roster. This can lead to more variation in the week’s work between rosters. However, since there is more stability in each driver’s roster, this may be preferable.

 

Minimize variance between work-weeks

Here it is desirable to equalize the total work time between rosters. This makes all the work done in a week as equitable as possible for all rosters, but may cause more variance between start time on each day of a roster. Although this makes the rosters as "fair” as possible from a total work perspective, it may decrease the driver convenience. Again, in some situations, this may be preferable.

Blend driver convenience and minimized work variance

The ultimate solution would be to be able to choose between convenience and equality, and The Master Scheduler allows this too. The scheduler can choose along a sliding scale to indicate the level of preference for driver convenience or equality. This feature is only available in The Master Scheduler software.

Conclusion

The Master Scheduler supports a wide range of techniques for rostering to meet the needs of many properties. By allowing the scheduler to choose the technique dynamically, the software allows "what-if” scenarios to be tested quickly without the need of costly additional consulting. This ensures that schedulers using The Master Scheduler will continue to meet the rostering needs of their property today as well as in the future.