Introduction
This is the third and final part of the series on bus scheduling. The first two parts can be found here.
Remember that OC Transpo claims that their system "permit[s] more flexibility," "increase[s] efficiency," and "reduce[s] operating costs."
In fact, their scheduling system is not flexible, does not increase efficiency, and increases operating subsidies through a combination of unreliable service and increased deadhead and layover times, which are not, or cannot, be manually revised to ensure cost savings. By ending random interlining and looking at bus operating cost from a standpoint of bus allocation rather than the endlessly stupid "1 minute of travel time savings is $40,000 saved!" from the last Transit Committee meeting, we can improve service, if only the network and schedule design was treated as part of the process rather than as stone tablets handed down from God himself.
The scheduling process for a spring service period starts the previous October or November, when the service change details are finalized by the network service design team. These details are handed over to the scheduling team in mid- to late-December. With this information in hand, the scheduling team can start their work for a spring service change by the first week of January.
During the first half of the scheduling cycle, the scheduling team inputs the changes received from the network service design team, which typically includes details such as route changes, service level adjustments, special event service details, and construction and detour information. This information is used to create timetables for each route, which contains information about trip times and the span of service from the first trip in the morning until the last trip of the day.
During the second half of the six-week scheduling cycle, the primary task is to create the schedules for bus operators, electric rail operators, and diesel rail operators ... [editorial note: OC Transpo uses HASTUS software to automatically assign both vehicles and operators. The process of assigning operators is usually automated in other cities as well, but to my understanding, that is not the norm automated for vehicle assignments.]
During this period there is not a lot of flexibility for iterative reviews or adjustments after the service changes have been input into the system. It is important to have all the details finalized before this period starts to minimize the need to make changes and rerun the scheduling software that is used to generate and optimize the schedules. The inputs to the scheduling software are very complex and contain large amounts of data and details, and can take several hours, or even days, to run on the very fastest computers the City has.
Case Studies
A few months ago, I wrote about layovers being scheduled in terms of "minutes per hour." We're going to come back and look at some examples of long layovers, and how they could be flattened to improve service.
The first examples, which I am bundling together, are 4797 and 6508, both of which spent substantial time operating on Route 45. All of these examples come from 8 May, but are relevant to the entire Spring service period and generalizable to every service period, even if the exact operating combinations change.
Both buses operated on many routes, and there are many things wrong with these blocks. For the purposes of this piece, we will focus on Route 45.
The layover time when the buses operate on Route 45 is about 30 minutes/hour. Route 45 is only a short shuttle between Hurdman and the TOH campus on Smyth Road; its scheduled roundtrip time is about 15 minutes long, but in practice it turns out to be closer to 16-17 minutes. This route operates frequently, so it requires two buses to run every 15 minutes and one bus to operate every 30 minutes, at current standards.
I have provided a string diagram of this route's operations below. Buses are generally assigned to this route for long periods at a time; they do not just interline with another, less frequent route with an awkward layover at Hurdman, which would be both convenient and highly implausible.
There are two solutions. The first is to cut runtimes to reduce the trip requirement to 12-13 minutes (formally, the runtime could be set to 14 minutes, but that is not feasible in the real world). This does become a solution on other routes, which are explained below.
The other solution is to improve service, using the existing bus allocation. Instead of every 15 minutes, the <45> could be run every 10 minutes, and there would remain 15 minutes/hour of layover time, which is more than enough. The bus fleet required is two buses, which would not change the cost of operating the service, except for diesel fuel running two extra trips an hour - and this cost is offset by increased ridership from improved service. When service currently operates every 30 minutes, this could be revised to run every 20 minutes at no cost.
The schedule isn't the only thing that needs to be treated as part of the network. In some places, route design needs to change to take full advantage of the fleet - and strategic interlining could be part of that.
An example of both is bus 4777.
Both Routes 165 and 162 show ridiculous layover times. In the early morning, route 165 has a scheduled 25 minutes/hour of layover; the 162 is scheduled for 30 minutes/hour, but realistically sees around 28 minutes/hour of layover time.
In the midday, both routes run with one bus each. This means we are paying two drivers to sit at Terry Fox and not run service.
Unlike with Route 45, these routes cannot simply get a schedule adjustment and be on their way, as neither would reach the minimum 5% layover requirement, and running unrealistic schedules is not desirable to reach theoretical efficiency outcomes.
Here, the solution is to modify the routes to run a shorter runtime. These two routes could be interlined through Terry Fox, in which case about ten minutes of combined runtime savings need to be found to run half hourly service in the midday. My suggestion is to run the 165 on Kanata Avenue instead of Terry Fox, and the 162 bypass Arcadia. However, all solutions must pass political interference from Council, who micromanage the transit network while neglecting the overall picture. In this case, alternatives should be presented to residents and council (including the status quo), which utilize these buses better than the current situation.
Once that runtime saving is achieved, routes 162 and 165 can be merged into one route (or split, with an informal but scheduled interline through Terry Fox) that use two buses on a half hourly headway, thus doubling service for two routes at no cost. The only change needed is to the route network, after an end to random interlining.
The other side of this change is that once set down, the route must stay at a consistent runtime. Constant network changes are a negative on ridership, so the City must commit to transit priority and OC Transpo must commit to best practices to ensure routes run with the same travel time even as congestion increases in growing areas such as Kanata.
Bus routes with short-turn trips can also use this kind of thinking and creativity.
Bus 6527 operated on the 39 for much of the day: 
Route 39 is currently operated so that the half hourly Blair-Millennium service requires three buses, with a layover of about 9 minutes/hour, while the short-turn Blair-Trim service requires two additional buses with a layover of 22 minutes/hour. This pattern can be seen in the string diagram below.
The goal is to improve the utilization of the two Blair-Trim short turn buses. The first is to extend frequent service to Millennium, which would provide frequent service to a substantial number of people in east Orleans at the cost of one extra bus. Another would be to run service every ten minutes between Place d'Orleans and Blair with these two buses; this would, however, mean an end to frequent bus service at Trim (which is likely a moot discussion with Stage 2 East coming online soon anyways). These changes would need to be evaluated for feasibility, cost, or ridership at Trim, but represent ways to improve the service and efficiency of OC Transpo for low costs.
In any case, this kind of thinking should be applied to every route, short-turn or no. OC Transpo should plan for bus allocation, not for cutting runtimes by one minute, as that has no effect in most cases. The network and its schedule is an integral part of cost planning, and need to be considered as such.
The final example represents another example of schedule unimaginativeness - route 93, which is a shuttle in Findlay Creek. This example uses bus 4696.
Route 93 is an isolated route, which interchanges with Line 2 at Leitrim Station, the only substantial transfer on the route. An edited map is included below.
The present arrangement is that two buses run on a half hourly basis. The route has a runtime of 30 minutes/hour, so the layover is 30 minutes/hour. Moreover, since Line 2 runs every twelve minutes, transfers at Leitrim Station do not correspond to Line 2 schedules, resulting in long and unpredictable wait times for riders. -edited-
A fix for this problems requires both a schedule change and some concrete.
Route 93 needs to be revised to run at multiples of 12; probably every 24 minutes, which would allow it to run with the same cost as today. Conveniently, NB trains towards downtown arrive at Leitrim Station two minutes before SB trains from downtown; the bus should be adjusted to provide a timed transfer, arriving at Leitrim Station 3-4 minutes (for schedule fragility reasons) before the northbound train departs and leaving Leitrim 2 minutes after the southbound train arrives. This provides a layover of about eight minutes; the other eight minutes in the schedule need to be burned at the other end of the route, necessitating either a new turn around and layover facility on Bank or Rotary Way, as well as a change to run this route on a two-way basis its entire length.
Two buses would allow for meets every 24 minutes, connecting to every second train. A three bus allocation barely misses the requirements for the eight minute layover while meeting every train, but minor changes would be enough to make this a reality - a better signal at Leitrim/Gilligan and bypassing the Kelly Farm/Bradwell detour would likely be enough to bring runtimes for this to be feasible. It would also avoid the need for layovers in Findlay Creek, at the cost of adding a bus (but also much ridership on what would be a very attractive service) to the route.
This change would create what is known as a knoten, in German, or a pulse in English-speaking planning, at Leitrim Station, to which more routes could be added, creating convenient transfers not just between trains, but also between local bus routes such as an improved route 94, or new routes towards Osgoode and Rideau-Carleton Raceway.
Conclusion
To hear them tell it, OC Transpo's scheduling practices are innovative, with major cost savings while ensuring service is provided. These practices are uncommon for a very good reason; it increases costs through overuse of interlining and encouraging inefficient route and schedule practices by staff; it reduces flexibility by setting "cheap wins" out of place, since that extra trip will be gone next service period; and it turns schedules into a Frankenstein, unable to deal with delays, extended travel times, and totally fragile and totally vulnerable to disruptions.
Schedule reform should be a top priority for OC Transpo management, or for Transit Committee (who are aware that there is a common sentiment OC Transpo is unreliable, at the very least). City Council needs to stop micromanaging routing choices, and allow staff to set out route structures that take advantage of bus resources and allocations, and give them the tools to enforce travel times without knocking an entire route network out of shape.
In this series, I have only presented the most egregious examples of bad scheduling, visible to a non-professional eye. I have also proposed some solutions, which may be viable and may not be, in the context of what we need - a comprehensive review and reform, which would provide insights into opportunities across the network. Schedule reform would provide many benefits for a low cost (arguably a net positive on cashflow). It is the fastest, cheapest, most essential, and most effective improvement we could bring to hundreds of thousands of people who rely on bus service in Ottawa everyday.
It also provides a symbolic window into the City of Ottawa with OC Transpo's insistence on being a special case study for other cities to avoid, an example of the continuing neglect of OC Transpo management, and City Council's refusal to learn about an albatross of a transit network from other cities best practices, instead playing with minor changes around the edges that do not address the real challenges transit faces at all. Burying our heads in the sand won't improve our transit network - only action, learning, and political courage will.
Until next time.
No comments:
Post a Comment