Click for next page ( 23


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 22
Chapter 2. Inputs to the Scheduling Process sign-up Annual Service Plan The process in which operators An annual Service Plan is not something that is uniformly prepared and followed, but it is a select work assignments. Most decided advantage to the scheduler. An annual service plan, which can exist as an Operating agencies have three or four sign- Plan or a Service Budget, takes the proposed service changes, additions and deletions and ups each year. incorporates them into a time framework that takes into account the sign-up process (when operators pick their work assignments). An ideal format for an Annual Service Budget, as sign-up period developed at one large transit agency, lists service additions, discontinuations and adjustments chronologically for each sign-up period during the upcoming year along with an estimate of The period of time that a specific sign-up is in effect, usually three or the vehicle hour impacts. It also carries a line item called "augmentation" which includes non- four months. allocated hours set aside for trip increases to correct overcrowding on peak-only routes, such as park-and-ride express routes. The schedulers also have latitude to eliminate trips or peak hour buses from underperforming lines and use these resources to add where they are better vehicle hours needed. Seasoned schedulers will often refer to having adjustments "in their pocket," which Total hours of travel by a vehicle, means they already know where they can reallocate or cut hours if the budget requires it. including hours in revenue service (including layover time) and At most systems, the sign-up process occurs three to four times a year. The advantage of this deadhead travel. Also known as is two-fold; ( ) the scheduler can plan their work for a complete year and ( ) they can work "bus hours" for bus. "Car hours" is within very close estimates of what the annual budgeted operating hours and/or miles will be. the term used for rail. This becomes critical at a time when major service cuts must be made to meet unanticipated budget shortfalls. mode A type of transit service character- 2.4 Organization of Scheduling Departments ized by vehicle or operational features. Common transit modes A question that is often posed is "what is the ideal size and composition of a scheduling section include motorbus, trolleybus, light of a modern day transit system?" The answer requires a great deal of input information and rail, heavy rail, commuter rail, and includes some of the following criteria: demand-response. How often does the service change/How many bids per year? How much service changes each time? exception scheduling Are there multiple modes? Scheduling activity undertaken to Does exception scheduling account for significant amount of scheduling time? address major construction detours or delays, sporting events, holiday How large are the garages and how many are there? service, or other special situations. To what extent are schedulers required to undertake service planning work? Do schedulers prepare rosters? How large is the transit system (the schedulers-to-bus ratio is not simply a straight line function)? 2-13

OCR for page 22
Chapter 2. Inputs to the Scheduling Process To what extent are schedulers responsible for maintaining data for downstream sys- tems? How complex or constrained are the work rules? In short, the factors that define scheduling department size are far too numerous to allow a simple standard. If there is truly an average out there, it is one scheduler per operating garage. A garage on a medium to large transit system can store anywhere between and buses, with the average being between and buses. Schedulers may employ a few assistants. The scheduling department will invariably include staff with a range of experience and skills. As a result, a natural experience-based hierarchy can be derived. In addition to schedulers there may be scheduler trainees or interns, scheduling analysts or similarly named positions, one or two senior schedulers, and the head of the section. Additional support staff may be needed to perform a growing number of ancillary functions, such as programming headsigns and auto- matic stop annunciator systems and keeping up the bus stop list. At some systems (typically smaller agencies), schedulers are responsible for carrying the whole task of building a new schedule through from analyzing the traffic checks to the finished runcut and roster or bid package. At others, schedulers schedule and senior schedulers perform the runcutting, which is considered to require the work of a more highly skilled person. The advent of computerized scheduling tools and even of general analytical tools such as spreadsheets has changed the nature of scheduling departments. These systems tend to allow more interaction among the traditional scheduling processes (schedule writing, blocking, runcutting, and rostering), reducing the tendency for schedulers to focus on only one aspect of the process. Our experience is that it is more difficult to achieve proficiency in building tightly constructed schedules than in other areas of scheduling. There is almost an infinite number of ways that revenue service can be scheduled, which means this proficiency takes much longer to learn and comes easier for people with a certain set of intellectual aptitudes. revenue service Scheduling Data Sources When a vehicle is in operation along Historically, schedule data was collected by real people recording passenger and time informa- a route and is available to the public. tion on pieces of paper. This "traffic checking force" was a critical building block in scheduling. Now,many systems use their traffic checkers primarily to gather data for federally mandated reports (such as the National Transit Database) that are of little or no use by scheduling staff. Therefore, traffic checkers may report to a service planning section rather than to the schedul- ing department. 2-14

OCR for page 22
Chapter 2. Inputs to the Scheduling Process point checks Data collection for scheduling can be conducted using traffic checkers, APCs, or AVL systems. A technique to collect information Human checkers are still needed for special checks, e.g., at special events (concerts, sporting about passenger loads and schedule venues) and on rail systems where per-car loadings are required. adherence at a single location (or Two types of traffic checks are specific to scheduling and are vital for obtaining the information point), typically a time point or a necessary for adjusting an existing schedule or building a new schedule. These are generally location where branches of a route diverge. Also known as "line check." referred to as "point checks" and "ride checks." Note that even for data collection with human checkers, the use of technologies is an integral ride checks part of the process. Almost all counts use spreadsheet or database systems for production of checker sheets, data entry, and analysis. Handheld devices can be used to undertake the actual A technique to collect information about boarding and alighting at collection, with a range of obvious benefits. every stop, in addition to passenger Point Checks loads and schedule adherence at all time points. Ride checks may also These are the simplest and take the fewest number of people to perform. They are most ap- include data collection on type of fare plicable for schedules that are demand-driven, rather than policy-based. A checker is stationed paid, stop announcements, or other at a point on the route that is known to be the place where the greatest number of people are information of interest to the agency. consistently on board. This point is known as the maximum load point or MLP for short. The Ride checks are more labor-intensive checker records the number of passengers on board the bus either arriving at the stop, leaving than point checks, but provide more the stop, or both, along with the arrival and/or departure time. Other identifying information, complete data for a given route. such as the bus number and block number are recorded. General information about condi- tions is also noted, such as the weather and any unusual traffic conditions. maximum load point Point checks may also take on the nature of a spot check where unusual conditions occur, such The location along the route where as boardings at schools at a particular time of day. These are short in duration, just covering the the passenger load is greatest. The times when boardings are the highest. They may be made by scheduling personnel who need maximum load point can differ by the information quickly to correct a capacity problem. Besides the MLP, point checks may be direction and by time of day. Long or regularly made at other points along the route, such as where branches diverge (or just beyond complex routes may have multiple the point of divergence), or where short turns are scheduled. This gives vital information about maximum load points, one for each the split of ridership along each branch. It is possible to obtain running time information by segment. staging checkers at each time point and comparing their times, but it is vital that all checkers synchronize their watches so the information coordinates correctly. block number On the subject of time, it is highly advisable to record both arrival and departure times, prefer- A unique number associated with a ably required to the half minute (handheld devices allow recording to the second). The reason specific block, used to track the block for this interest in half minutes (not that many systems still write schedules to this tolerance) is throughout the scheduling process to provide greater accuracy in compiling running time information. End-to-end running times and as a means of identification for are generally rounded to the whole minute, but the half minute can make a difference when the operations department. allocating times between time points. 2 At some systems, buses display run numbers instead of block numbers in their windshields. 2-15

OCR for page 22
Chapter 2. Inputs to the Scheduling Process Bear in mind that one day of data may not be sufficient on which to base a schedule adjust- average weekday ment. There are too many variables in the operating day to be able to rely on one check as A representative weekday in the a good "average" of how the route operates regularly. Schedulers are constantly trying to operation of the transit system construct the average schedule for the average passenger loadings operating in average street computed as the mathematical conditions and weather through average traffic. Since we know that the average day does not average of data for several typical really exist, it is important to have enough checks to be able to distill down to that average. We weekdays. A typical weekday is one hope we have made that point! where there are no anomalies such as high ridership due to extra service So, how many checks would yield a good representative of average conditions on a route? added for a special event or low Those of us who began in the pre-computer days were taught to get counts on three days and ridership due to inclement weather. compare them. Did they look similar? If so, take an average weekday. If not, throw out the Some schedulers claim that this does outlier (i.e., the significantly dissimilar observation). Today, with greater data availability, the not really exist. Average Saturday answer is really a function of the variability in loading and operation from day to day. Each and average Sunday are determined route will have its own variability in these factors. By taking a few samples over time, use of in the same way. sampling methodologies can easily be applied to identify confidence level estimates for this variation. Farebox data can provide a good basis for estimating sample requirements--simply measure the variability of ridership across days at whatever level of aggregation is required (daily, time period, or even trip level). This can tell you how big a sample is required. The following is an example of a point check. These should be developed in a spreadsheet or database, which is also then used as the basis for data entry and analysis. This ensures consis- tency and improves accuracy. 2-16

OCR for page 22
Chapter 2. Inputs to the Scheduling Process 2-17

OCR for page 22
Chapter 2. Inputs to the Scheduling Process On-Board Ride Checks If point checks seem to be so ideal, why do anything else? Point checks generate only a subset of the information needed to undertake many scheduling tasks. Their key benefit is that they collect a good deal of what is required with minimal manpower. One of their greatest deficien- cies is they do not tell the scheduler anything about what is happening at other points along the route, but only provide a snapshot of what is happening at one location. A far better solution is to collect full data for each trip through an on-board ride check. This is also more appropriate for policy headway routes, where patronage is lighter. It requires more resources but results in a more complete data set, providing the scheduler more information to make informed schedule adjustments. The systems that regularly perform ride checks tend to have, proportionate to their size, a far larger checking staff. Again the issue of sample size needs to be addressed for the ride check. In this case the sample size relates to what percentage of trips on a given day are included (whereas for point counts the question was how many days). A particular point needs to be made here that you prefer- ably check the entire day and not try to do half the trips one day and the rest on another. Days are unique. Loads are different. Operating irregularities are different. Two partial days of ride checks blended together may either accentuate problems or under emphasize them, but the blend does not always give you a representative look at the route. On-board ride checks are tabulated on forms which list all stops from end to end and provide columns for showing scheduled time and time points, for writing in the actual times and for noting comments about drivers' performance in operating over the route. These comments could address speed, unexpected stops/delays, and other aspects of interest to the agency. In addition, there are columns for recording boardings and alightings at each stop and carrying a running total of how many passengers are on board throughout the trip. 2-18

OCR for page 22
Chapter 2. Inputs to the Scheduling Process T R IP S U R V E Y S H E E T R oute/D irection: 70 S B C hecker N am e: A ssignm ent D ay of W eek: S urvey D ate: T rip #: R un # PERSONS PERSONS TOTAL ON LEAVE G E T T IN G G E T T IN G BUS W HEN T IM E STOP STOP NAM E OFF ON IT L E A V E S P O IN T 1 M o n teb ello /JC P en n ey#2 0 : 2 M arkland/P otrero G rande 3 M o n teb ello P lz/P arkin g L o t : 4 W ilcox/V ia P aseo 5 W ilcox/H ay 6 W ilcox/Lincoln 7 W ilcox/V ictoria 8 W ilco x/B everly : 9 W ilcox/M adison 10 W ilcox/W hittier 11 W hittier/C oncourse 12 G arfield /W h ittier : 13 G arfield/O lym pic 14 G arfield/F erguson 15 F lotilla/G arfield 16 M etro lin k S tatio n : 17 M ines/V ail 18 M ines/M aple 19 M in es/G reen w o o d 0 : 2-19

OCR for page 22
Chapter 2. Inputs to the Scheduling Process T R IP S U R V E Y S H E E T R oute/D irection: 70 S B C hecker N am e: S m ith A ssignm ent 702 D ay of W eek: W ED S urvey D ate: 20-O ct T rip #: 20 R un # 70-51 PERSONS PERSONS TOTAL ON LEAVE G E T T IN G G E T T IN G BUS W HEN T IM E STOP STOP NAM E OFF ON IT L E A V E S P O IN T 1 M o n teb ello /JC P en n ey#2 0 1 1 3 :3 5 P M 2 M arkland/P otrero G rande 1 3 M o n teb ello P lz/P arkin g L o t 6 7 3 :4 2 P M 4 W ilcox/V ia P aseo 1 6 5 W ilcox/H ay 6 12 6 W ilcox/Lincoln 12 7 W ilcox/V ictoria 12 8 W ilco x/B everly 2 10 3 :4 6 P M 9 W ilcox/M adison 10 10 W ilcox/W hittier 2 1 9 11 W hittier/C oncourse 5 4 12 G arfield /W h ittier 1 3 3 :5 0 P M 13 G arfield/O lym pic 1 2 14 G arfield/F erguson 2 15 F lotilla/G arfield 2 16 M etro lin k S tatio n 2 3 :5 5 P M 17 M ines/V ail 2 18 M ines/M aple 2 19 M in es/G reen w o o d 2 0 4 :0 1 P M 2-20

OCR for page 22
Chapter 2. Inputs to the Scheduling Process automatic passenger Some days or seasons are more or less ideal for collecting representative data. This will vary counting (APC) by system, weather, and the importance of school ridership in the overall operation. When APC systems count the number of school is a factor, in-session times when classes are stable give the most accurate picture for boardings and alightings at each school time service level requirements. These are usually the months of March, April, October, stop while also noting time, loca- and November. Those are also the best when weather is a big factor. Days of the week differ tion, and direction. Infrared beams as well. Mondays and Fridays are not ideal checking days, as loads tend to be lower on those are the most common means days, and traffic on Fridays is different from other weekdays. Obviously, days before and after used in counting. Stop location is holidays are not typical weekdays either. identified through the use of data sources such as global positioning APC and AVL Systems systems (GPS), signpost emitters, APCs provide an automated version of on-board ride checks. The APC will collect time, loca- GIS maps, odometer readings, and tion, and passenger boarding/alighting information, much in the way an on-board survey does. inertial navigation. Data from all The obvious benefit of APCs is the capacity to generate many days of data for as many routes these sources must be extensively as the number of APC units will permit. The difficulties are in initial calibration of the systems, compiled (from multiple buses/trips in daily vehicle assignment (if only a portion of all vehicles are equipped with APCs), and in on a route) and processed, staff 's ability to analyze the sheer volume of data produced. either by an on-board computer or centrally, to be meaningful. The latter problem is a good one as far as scheduling is concerned. Schedulers work well with a wealth of data. Two common elements contributing to successful implementation of APCs are a sound validation program to flag data of questionable quality and good reporting capabili- automatic vehicle ties. The latter can include standard reports to provide schedulers with the information they location (AVL) need most often (boardings, alightings, loads and time by stop, and running time by segment) AVL systems are vehicle tracking and the flexibility to create ad-hoc reports to query the APC database as needed. systems that function by measur- ing the real-time position of each AVL systems track vehicle location throughout the day, but do not collect ridership data. Thus, vehicle and relaying this information the primary use of AVL data for scheduling is in the evaluation of schedule adherence and back to a central location. The running time analysis. As with APCs, data validation and reporting capabilities enhance the vehicle location is identified through reliability and usefulness of AVL data. Agencies that use both APCs and AVL tend to rely on the the use of global positioning AVL system for time-related data. systems (GPS). The information is used to assist transit dispatchers As more and more transit agencies acquire APC and AVL capabilities, either as stand-alone as well as inform travelers of bus systems or as part of a broader Intelligent Transportation Systems procurement, schedulers status. AVL is a potential source of will have more and better quality data at their disposal. APC and AVL will largely replace hu- running time and on time perfor- man checkers but, as noted earlier, there are instances where human checkers are still needed mance data for scheduling, but only (for example, cross-checking APC and AVL systems on buses not equipped with APC). A final if an archival reporting system is point is that automated data collection can tell you what is happening but cannot tell you why. included. Observations from operations personnel and fieldwork by the schedulers will remain an impor- tant part of the scheduling process. 2-21

OCR for page 22
Chapter 2. Inputs to the Scheduling Process We mentioned at the beginning that there were other types of checks that were less useful for schedulers. To complete our picture of data accumulation methods, we list these here: Trail checks entail following buses in a car over specific segments of a route. They can be useful for looking at general operations along a segment of route that is being studied for rerouting or discontinuation, but are not the preferred method for obtaining board- ing, alighting, or running time data. Farebox counts do give a record of how many patrons boarded on each trip (if the driver has been diligent to reset the box at the start of each trip), but they do not tell where passengers got on or off the bus. More importantly, fareboxes cannot tell the maximum number of passengers on board for each trip. Farebox counts are useful for identifying ridership trends. Cordon counts are a common means in traffic engineering to count vehicles and people entering and leaving a specific area, typically a city center. Some of this information at a specific corridor level may be of interest to transit, but overall results are of little use for scheduling purposes. 2-22