✓

The Case
for ROSTER
OPTIMIZATION
 ✏

Introduction

Bus operations are complex, and so is scheduling. It is a combination of craft

and science, and transportation operators devote considerable energy to getting
vehicle and crew scheduling (aka blocking and runcutting) right. They do this to
ensure efficient operations, cost savings, on-time performance and regulatory
(and labor rule) compliance.

Yet, the same amount of energy isn’t always applied to rostering. In fact, 71% of
US agencies surveyed by Optibus did not use roster optimization. Historically, the
industry has focused on using optimization during the scheduling phase and less
so during the rostering phase. Now, as technology advances, it is time to look at
rostering, especially since it impacts costs, driver satisfaction and safety. This
ebook will explain why and how.

1
What is rostering?
Rostering is the last optimization step. First come route planning and timetables.
Then vehicle and crew scheduling (blocking and runcutting) as well as relief
vehicle scheduling. Rosters are last.

Rostering is the process of grouping daily operator runs (shifts or duties) into
packages of weekly work assignments. These assignments are typically five
eight-hour runs or ten four-hour runs. They may include a combination of
different runs such as: weekdays only, weekdays and Sunday/Saturday, and other
combinations.

Rostering can be done in several methods. They include manual rostering by

schedulers; “cafeteria-style rostering” in which drivers build their own rosters
by picking which runs and days off they want; and the use of roster optimization
algorithms through a planning and scheduling platform.

Of all three methods, only roster optimization uses technology rather than human
input to ensure the creation of the most high-quality rosters while reducing
overtime costs, adhering to break rules and rest time, and implementing driver
preferences (for instance, regarding split shifts or start times).

2
Why is roster
optimization
important?
Overtime pay and guaranteed pay:
Since (in most cases) overtime is determined on a weekly basis, rosters are the
only way to control overtime costs, which can represent 2-5% of all labor-related
costs. Reducing the ratio of overtime to total paid time can offer significant
savings. Similarly, when drivers are guaranteed pay if they work less than a
minimum amount of hours, roster optimization can be used to ensure they work
these hours rather than be paid for guaranteed time. Roster optimization does
what the human cannot: even out duties into packages that will contain the right
amount of weekly hours for all drivers, without having too many drivers that get
too many or too few hours.

Fairness and delivering quality rosters for

drivers:
The ability to spread duties in a way that will provide good outcomes for all
drivers, instead of the lucky few that got overtime or a full 40-hour-a-week
roster, provides a fairer outcome for drivers. Fairness is an important source
of emotional satisfaction at work, and can help retain drivers and even provide
hiring incentives. Roster optimization means there are many good options for
many drivers, ensuring more drivers are happy with their roster assignments and
overtime is allocated only when needed.

Some agencies allow their operators the convenience of "cafeteria-style rostering,"

meaning operators can pick their schedules. They get to pick their shifts a day at
a time, and those shifts add up to a roster. Normally the order of picking is based
on seniority. Many times cafeteria style-rostering creates a duty allocation that is
not equitable nor efficienct.

3
Reducing the roster count:
In some cases, optimized rostering can reduce the roster count, meaning that
fewer drivers are needed to cover the same routes. This can be helpful in times of
driver shortages and can save costs. Schedulers can constrain roster counts as
part of a what-if scenario to see whether they are better off with more rosters and
less overtime, or whether overtime with a reduced roster count is better.

Compliance and safety:

Roster optimization makes it easy to comply with labor laws, union agreements
and safety requirements such as rest time between duties and breaks during
duties. This ensures all rosters are compliant and valid and saves valuable
scheduling time.

Dealing with different driver groups:

In case the transportation organization has different drivers to whom different
labor rules apply, roster optimization is sometimes the only effective way to
make sure the organization is complying with the rules for each group, and such
optimization can produce valuable savings.

4
Creating rosters
Three key components factor into creating a viable roster:

Compliance with regulations & collective

bargaining agreements (CBAs)
CBAs and other regulations determine the minimum/maximum weekly (or period)
hours, number of workdays, time off requirements, break regulations, etc. All these
rules and preferences (for instance, at least two consecutive days off) need to be
reflected in the roster or schedule.

Safety requirements
Safety is crucial when creating a roster; making sure drivers get enough rest
before their shift and that shifts are spaced correctly can be an important safety
requirement. Other requirements relate to duration of duties, start and end
times, length of the duties prior to a break, etc. In some countries, transportation
regulators are attempting to create fatigue indices to model fatigue and ensure
operators can drive safely.

Driver satisfaction
Roster optimization isn’t just about ensuring savings on overtime and guaranteed
time. Rosters are an incomparable opportunity to improve drivers’ work-life
balance. This can be anything from route familiarity, similar start time and end
times, and rosters that are designed for specific groups, such as retirees with a
part-time job or single parents.

Think of this as a way of aggregating driver preferences (morning shift/late shift,

start times, routes, etc.) and taking them into consideration to create as many
quality roster packages as are required. This can help deal with driver shortages,
prevent burnout and offer good conditions for new drivers that join. Before you
begin rostering, make sure to interview drivers and check with them what is a
good roster from their point of view.

5
Roster optimization
in practice
Here are some practices that advanced scheduling platforms offer schedulers to
help them create better rosters.

Comparing roster scenarios

In scheduling, there is a simple truth: more scenarios equal better results. Using
quick and powerful optimization can assist in creating many roster scenarios
quickly. This enables a better understanding of the trade-offs associated with
different roster rules.

Here are four KPIs you should pay attention to in each scenario:
• Overtime
• Make up (guaranteed) time
• Pay hours
• Number of rosters

Four or five-day workweek

Probably the first thing to check is the impact of a four-day workweek with 10
hours a day, a five-day workweek with eight hours a day, and a mixture of the
two. This can be easily defined in the roster rules and preferences settings, and
scenarios can be prepared to check the options and their impact on business
KPIs. Additionally, different “roster groups” can be created with some drivers in
the 4-day workweek group and some in the five-day workweek group.

Consecutive days off

One of the greatest benefits for drivers, regardless of which work week they have,
is to make sure that days off are consecutive, so that their time off works well
for them. Additionally, you can use roster optimization to put short shifts (if they
exist) just before days off, so that the time off is long and uninterrupted.

Start times and route variation

The following example is a scenario focused on overtime cost reduction. This
looks like a successful optimization: The result is zero overtime, meaning in this
case that about 1,500 hours are saved annually. Yet, this is no success if the
rosters are of low quality for drivers.

6
 KPI Current Schedule Scenario 1
Vehicles (PVR) 55 55
Roster Count 115 115
Over Time Hours 58 0

In this case, low quality can be related to two issues: too many routes per roster
or a large variation in different start times for each day, which can be a problem
for drivers. This can be addressed by modifying the roster preferences, such as
requiring similar start times or limiting route variation. This is also an example of
where scenario analysis can help, because we can test several requirements for
route homogeneity and see which works best.

Evaluating different scenarios

Take a look at the following scenarios. Each one of them has a different goal. One
is focused on zero overtime, another on reducing the number of rosters, and the
third on start time similarity.

KPI Scenario 1 Scenario 2 Scenario 3

Minimum Number Start time
Scenario zero overtime
of Rosters similarity
Number of Rosters 144 130 133
Weekly Hours 5312 5312 5312
Over Time Hours 0 119 160
Over Time % 0% 2.2% 3%
% of Rosters with 65% 75% 92%
90min start time
difference
# of 4-day Rosters 134 76 68
# of 5-day Rosters 4 54 57
Part-Time Rosters 6 0 8

Scenario 1 does not allow overtime, but the trade-off is that only 65% of the
rosters have a similar start time, meaning that there is no overtime, but perhaps
the overall roster quality is lower, from the drivers’ point of view.

7
Scenario 2 focuses on reducing the number of rosters, reducing roster count by
14. Yet it adds 119 overtime hours, which may still be plausible given that overtime
here is perhaps required to deal with driver shortages.

Scenario 3 focuses on enabling drivers to start their shifts at a similar time each
working day. Here, the trade-off was a 3% overtime increase, adding 41 overtime hours.

Roster groups
Roster groups enable the creation of different types of rosters for different groups
of drivers (for example, new drivers vs. veteran drivers). Veteran drivers may enjoy
rosters that suit their preferences (reflecting their tenure). It may make sense
to create different groups for part-time drivers and full-time drivers or create
different groups for drivers with a four-day workweek and drivers with a five-day
workweek, or different groups for drivers to whom different union rules apply.

Using advanced scheduling platforms allows schedulers to easily input different

driver preferences and cater to a variety of driver needs within the specific roster
group, while ensuring optimized results for each group separately and across the
different groups.

Rotating Rosters
Rotating rosters mean rotating duties between the different members of the
roster. A simplified example of such a roster would be a rotating roster with 3
drivers: Tim, Jane and Bob. Tim is assigned roster line 1, Jane is assigned roster
line 2 and Bob is assigned roster line 3. Since each driver moves on to the next
roster line after each is completed, the next week Tim would get roster line 2,
Jane would get roster line 3 and Bob would get roster line 1.

8
There are two key benefits to rotating rosters.

✓ Equitable duty allocation for drivers

Assuming roster line 1 is the most preferred, only one driver gets it and the
rest have to do without it. In rotating rosters all drivers in the group will get the
preferred roster lines at least some of the time.

The roster lines for a given set of drivers can be similar (for instance, they can
have roughly the same start time), or operators can cycle through alternating
roster lines (for instance, an early week followed by a late week). Either way,
drivers with rotating rosters benefit from knowing what their week will look like far
in advance, allowing them to make personal plans around their working hours.

In addition, some operators say they prefer taking on the diversity of tasks
that may be available with a rotating roster because it makes their work more
interesting.

✓ Maintenance becomes easier

Since all the operators within the group are familiar with all the duties, It is easier
for operators to cover for each other if someone is sick or running late. This duty
familiarity makes it easier for dispatchers as well. With rotating rosters, drivers
can still get “good” roster lines, while schedulers don’t have to work hard on
tailoring them to specific drivers.

Validations
Last but not least, advanced platforms offer schedulers the benefits of using the
power of algorithms to validate the rosters and make sure they comply with all
necessary labor agreements or alert the scheduler about any violations.

When a non-rotating roster is optimized, each of the rules is validated within a

given roster line, since operators perform the same duties each week. When
rotating rosters are optimized, the validations take the adjacent roster line into
account as well, since drivers switch duties from one roster line to the next.

In some cases, duties that are valid when looking at one roster line at a time are
invalid when evaluated in the context of a rotating roster and vice versa due to the
driver change that happens in a rotating roster.

9
Summary
It’s time that the technology allowing roster optimization be put into practice.
Roster optimization is a win-win for all involved. It saves time for schedulers,
freeing them to focus on passenger satisfaction; it offers better conditions for
drivers, ensures compliance with regulations and union rules, and saves money.

10