In Part 1, we discussed the importance of schedule adherence/on-time performance (OTP). We discussed the institutional aspects, such as appropriate measurements and the impact of policies — coverage, service levels, and achievability. Most importantly, we presented the concept of using low OTP as a call to action, not just a number to report.
We introduced the concept of the “Five Musts” — moving from institutional analysis and concepts to implementation. In this part, we’ll discuss aspects of these requirements and suggest approaches.
The MUSTS are:
- Adequate and accurate running times between timepoints (including adequate pullout deadhead times;
- Adequate recovery time;
- Enough trained, qualified vehicle operators;
- Enough equipment (transit vehicles) available for use;
- Enough trained supervisors to monitor and adjust service as necessary.
1. Adequate/Accurate Running and Deadhead Times
As we saw in Part 1, nationally reported revenue speed for normal transit service (excluding commuter and bus rapid transit) has decreased by 8.4% in the last 22 years, from an average of 13.09 to 11.99 mph. While some transit authorities have overhauled their systems, often shifting from a “coverage” to a “frequency” philosophy, many are still operating routes and schedules from decades ago.
In much of the country, lower-cost land has moved demand generators farther and farther out, extending routes at a time of increased congestion and shrinking speeds. Often, schedule changes haven’t kept pace with these two factors, so the third leg of the coverage-frequency-service-quality triangle has suffered — poor OTP.
National speed averages derived from the National Transit Database have limited local value.
- A route in Upstate New York was unsustainable at a scheduled speed of 7.5 mph since it connected senior ctizen complexes with shopping centers.
- A route in the upper Midwest could operate faster than 15 mph because there was a stretch of State highway with no stops.
- An urban circulator operates at less than 6 mph, while a southern system operates two divisions where routes can range between 10.9 and 13.7 overall mph, due to different area characteristics.
Overall route speeds in your city are only useful for general planning — feasibility purposes. When a new route is being designed or an alteration is under study, using your system’s average mph helps predict whether the change may work.
Quality service requires focus on the details — running time between timepoints. Part 1 discussed a system with too many timepoints, some as close as one, two, or three minutes apart. For various reasons, they were unhappy with their OTP. A trail-check found buses passing the first few timepoints early, since the operator knew that running time would be required later. When timepoints are too close together, it’s virtually impossible to predict running times with sufficient accuracy and granularity — the reason that best practices say that timepoints should be about ten minutes apart.
Every route segment — between timepoints — is likely to have different operating characteristics.
- The way to accurately predict running time between timepoints is to simulate service, including mock passenger stops, using the type of vehicle which will be operated.
- When possible, use an operator from the middle of the seniority list. Very senior, experienced operators know how to run on time and make it look easy, while operators, just out of training, lack experience.
- In considering an existing route, avoid using pure data, such as that generated by AVL/APC’s. While that data — especially the “mode” — the number that appears most often — is valuable, it must be tempered with actual observation.
- If there is too much running time, experienced operators will “drag the line” or operate slowly, so the computer won’t pick up any discrepancy.
- On the other hand, when time is insufficient, may operators will speed and take risks to operate to schedule and serve their passengers. In my experience riding with operators who run red traffic signals — the common factor was tight schedules.
While mph statistics should not be used to predict running times — actual observations are best — they are useful as a “sanity test.” Scheduling software isn’t necessary, a simple Excel file will work. Table 1 is an example from an existing system — anonymized by changing timepoint names. Obviously, this doesn’t meet our 10-minute timepoint guideline, but focuses on demand generators (Heavenly Hills is a subdivision turnaround, the rest are self-explanatory). The important element is that we can use formulas to calculate mph by segment, and consider the road types — and time of day — to determine whether our times look reasonable and consistent with other routes and segments.
Table 1 Example of Headway Sheet with mph calculations
There is an easy way to establish route-segment mileages. Many smartphone applications are available, particularly those for the running community. Open the application, drive the segment in any vehicle, and record the distance. Tenths of miles provide sufficient granularity.
Now, to ensuring we have adequate pullout deadhead times. If we don’t begin the day on time, the rest of the day suffers. Deadhead time is a cost. In contracted operations, it may be a cost uncompensated to the contractor — thus a pure expense. OTP, safety, and customer service will improve when sufficient time — perhaps including a little recovery time — is allowed. The first trip of the day is crucial. One large city improved OTP by focusing on the first trip of the day. While buses were leaving the garage on time, and deadhead was accurate, they discovered a phenomenon — the “breakfast sandwich.” Operators were stopping early in the day to make purchases — delaying service.
2. Adequate Recovery Time
Recovery time is the shock-absorber for anomalies that occur along the route. The old industry standard was 10%, or about six minutes per hour. In 1947, the American Transit Association, predecessor of today’s APTA, published “Bus Scheduling Manual: Traffic Checking and Schedule Preparation” — reprinted in 1982 as UMTA Report DOT-I-82-23. At that time, a contributing city used 10% for routes with round trip times of 40 to 45 minutes; 7% for longer routes.
In recent decades, however, that 10% has been found insufficient. Three prime factors include use of mobility devices (wheelchair lifts/ramps), bicycle racks, and random effects of increasing congestion. Today, many properties have found 15%, or nine minutes per operating hour, are needed. Interestingly, a study prepared by the University of Southern California “Optimal Slack Time for Schedule-Based Transit Operation,” used statistical analysis of a large transit property to determine the “right” number. They came up with 15% as being statistically valid.
3. Enough Vehicle Operators
There is a nationwide shortage of CDL vehicle operators, affecting transit as well as the truck and over-the-road bus industry. Solutions are well beyond the scope of any OTP study. That said, there are techniques that industry, including the transit industry, could consider. Today’s Railways Europe magazine (issue 279, March 2019) reports that Deutche Bahn, the German railway, is partnering with the country’s Federal Labor Agency in a pilot program to recruit and train refugees. Is there money available to you for such outreach? A system in Georgia hires operators as young as 18, trying to get them as they leave high school. Can you partner with the local school system — perhaps participate in a “Career Day"? Another possibility is contacting local truck driving schools — where transit’s selling point is that employees are home every night.
The operator work schedule — the runcut is a factor in the ratio of “good” to “bad” work. At least one transit system found that the traditional method, all the “bad” work being placed together, thus falling to the bottom of the seniority list, increased turnover. New operators were not willing to endure hardships long enough for work to improve. Hiring and training costs increased. Another technique, then, may be more appropriate. Mixing good and bad, a little “sweet” and “sour” in work that will fall to the bottom half of the seniority list may reduce turnover.
Appropriate use of scheduled overtime is also an appropriate technique. Managers may revolt at the concept of overtime. In fact, schedulers have often been directed to eliminate overtime due to the half-time pay premium expense. In many cases, this direction has come from managers (or accountants) unfamiliar with transit. If scheduled overtime is eliminated without a full complement of vehicle operators, the work either goes out at unscheduled overtime or does not go out at all — leaving stranded passengers.
A full consideration of the effect of overtime may change some minds. Using the example of 440 hours of work that needs coverage during the week, we see two simple options. One uses ten operators, each working 44 hours to cover the work. The extra four hours is at overtime rate which, of course, costs only the half-time pay premium. The net pay cost, then, is 460 pay hours. The other uses eleven operators, each working 40 hours. Total pay time of 440 hours seems to present a saving, until we consider the fringe benefit costs (and hiring/training) of that eleventh operator. There may be intangible benefits if the ten-operators-with-overtime scheme is used. Increased take-home pay improves morale. Missing work may cost the overtime premium for the week — reducing absenteeism. “Overtime is your friend.” On top of the mathematical calculation should be the overriding consideration. Elimination of overtime increases the number of operators needed — in an era when we can’t fill openings in the first place.
4. Enough Vehicles
This would seem to be a maintenance department function. However, there are actions that planning and scheduling can take to increase the effective number of vehicles. In operator scheduling, it is generally more efficient to replace operators at shift change times somewhere along the route — known as a “street relief.” The option is to replace not only the operator ending his/her day, but replace the bus with a fresh one — the “pull-out/pull-in relief.” That bus replacement generally adds cost — not only an added pre-trip/post-trip inspection, but added mileage-related variable costs. From a strictly scheduling, strictly operations perspective, the more costly option sounds like a bad idea.
If we examine the holistic needs of the system, though, we can use that increased cost as an investment in vehicle availability. An example could be a vehicle block out sixteen hours, which would normally operate with one bus and operators doing a "street relief." Instead, if we break the block midday and pullout a fresh bus, the maintenance department has one vehicle to PM in the morning (the afternoon bus), and can then PM the morning bus after it pulls in. If the transit system has a peak/base ratio greater than 1:1, this creates an effective increase in the available spare ratio — without costing a vehicle.
An additional benefit is that every such instance of beginning with a fresh bus/operator resets the clock and improves OTP. If the morning bus is late, the afternoon bus starts on time. Of course, this requires that the event occur at the end-of-line so passengers aren’t transferring midstream — but that should be self-evident.
5. Enough Trained, Qualified Supervisors
One school of thought is that automation can replace, or at least reduce the number of supervisors needed. Perhaps anyone mis-routed by GPS will question that idea. With appropriate training, dispatchers and supervisors can do more than respond to accidents and incidents. By actively re-spacing buses to maintain headways; by taking creative action to respond to service disruptions (turnbacks, discharge only, deadhead to on-time, deploy a strategic spare bus or use a bus coming out of service), they can have great effect on OTP.
Further, we should consider where our next generation of managers will come from. The lack of operators and mechanics in the workforce extends to a shortage of transit managers. Promotion from within, with appropriate training, ensures that we’ll have qualified managers — who understand transit from the ground up — for the future.
Let’s close with a few methods to increase velocity.
Rationalize the Number of Bus Stops
Bus stop spacing is a compromise between trying to make service convenient by reducing walking distance and maintaining velocity for passengers already on the bus. Perhaps the most extreme example of this is the “flag stop” system, still used on a small number of properties. It’s easy to conceive of how this may increase the number of actual stops — thus dwell time — as the bus stops every few driveways to pick up one passenger at a time. By the way, another problem with flag stops is that the boarding passenger now has an effect on safety — if they choose to board at a stop just over the crest of a hill or around a blind corner. The element to consider is that every stop to pick up or discharge a passenger, one at a time, lengthens the trip for remaining passengers.
Improve Operator Restroom Availability
Without facilities, and the time to use them, operators stop along the route delaying service. One west-coast system stated that they replace many seat cushions each year, unknown how many of those are in operator seats, as a result of urine soaking.
In 2015, a Congresswoman sent a letter to OSHA, requesting an investigation into the lack of transit operator restrooms. She cited health issues and mentioned a transit system that was ostensibly fined $3,500 for not providing restrooms but disciplining operators who delayed service by stopping to use the facilities.
Beyond the passenger delay and inconvenience of enroute stops, a safety hazard may be created by the unattended bus such as the one in the photo (shown left). The operator had parked the bus in front of a fast-food restaurant to use the facilities. The bus was rear-ended by an auto driver who claimed she didn’t see the bus because the sun was in her eyes. Fortunately, nobody was injured. The bus was destined for a light rail station about a mile away — where there are no restrooms. This discussion provides an opportunity for developing systems to balance the capital and operating cost of providing facilities against operator health issues, delays, and safety concerns that may arise due to lack of restroom availability.
Explore and Mitigate Fare Collection Delays
This may not be the most significant opportunity, but any improvement is a step forward. Streetcars were once designed with large front platforms so fares could be paid underway, or with circulation systems so passengers could pay a conductor as they passed mid-car. As farebox systems have become more complex, as bills replaced coins, transaction times have increased. In the electronic era, stored-value (tap) cards have improved boarding time in many cases. In others, though, delays result as passengers recharge the stored value cards. Safety generally precludes doing this while underway, since there is some interaction with or supervision by the vehicle operator. On-board sale of fare media — day passes and the like — also prompt delay.
Service quality, as captured by on-time performance, has a direct effect on passengers and may influence their choice of mode — thus transit usage. There are a number of considerations in how we evaluate our performance, and a number of steps we can take to improve the way we serve the public.
Arthur N. Gaudet is president of Arthur N. Gaudet & Assoc. Inc. (firstname.lastname@example.org). This article is based on his book, Improving On-Time Performance in the Transit Industry: A Practical Guidebook.