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 36
36
Base-file of 2003/04/10: 106 trips scheduled, 96 trips measured (91 %)
stop
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42
11:11
11:21
11:31 11:30
Tritapt 1.0 (b82) license holder is Peter Knoppers, Technische Universiteit Delft. Copyright © 1997-2006 TU Delft
11:41
11:51
trip MLT [hh:mm]
time [hh:mm]
12:00 12:00
12:10
12:20
12:30 12:30
12:40
12:50
13:00 13:00
1 3 5 7 9 11
0 2 4 6 8 10 12
distance [mile]
Source: Hermes (Eindhoven), generated by TriTAPT
Figure 5. Observed (solid) versus scheduled (dotted) trajectories.
determined and analyzed directly from the headway distribu- load relationship using a least-squared fit. Then, using that
tion. As an example, transit agencies in both Brussels and slope, Tri-Met normalizes the loads of individual trips to
Paris calculate, from headway data, the percentage of passen- what the loads would have been if the headway had been as
gers waiting longer than the scheduled headway plus 2 min. scheduled.
As part of this project, analyses of passenger waiting time
based on headway data were developed (see Chapter 6).
4.7 Demand Analysis
The sources of passenger use data are APCs and fare col-
4.6.4 Headway-Load Analysis
lection systems. In this report, the only fare records consid-
Headway and load have a simultaneous effect on each ered are location-stamped transactions, because analysis of
other--longer headways lead to larger loads, and larger loads farebox data at the route level or higher is routine.
lead to longer headways. Analyzing headway and load
together can lead to interesting insights.
4.7.1 Demand Along a Route
By analyzing headway and load together, Tri-Met created
a method for determining to what extent an overload is Passenger demand on a route, for a given direction and
caused by headway variation, as opposed to demand vari- period of the day, has three dimensions: geographic (i.e., along
ability (28). The idea behind this approach is that overloads the route); between scheduled trips (i.e., how is the 7:15 trip
caused by headway variation should be "cured" by better different from the 7:30 trip); and between days. Most analysis
headway control, while overloads that cannot be simply views aggregate over two dimensions and analyze the remain-
explained by headway variation may require a change in ing one; it is also possible to aggregate over only one dimen-
scheduled departure times or headways. Tri-Met estimates, sion, showing the other two in the analysis, as the examples
for a given route-direction-period, the slope of the headway- will show.
OCR for page 37
37
The geographic dimension of demand is shown in a vol- 4.7.2 Demand Across the Day and
ume profile or load profile, depending on whether results Scheduling Headway and
are expressed in passengers per hour (volume) or passengers Departure Time
per trip (load); another view shows ons and offs by stop.
By abstracting the geographical dimension using trip sum-
One graphical format, developed by Delft University
mary measures such as total boardings, maximum load, and
researchers and illustrated as the lower step line in Figure 6,
passenger-miles, one can focus on the other two dimensions
shows not only mean segment loads, but also mean offs, ons,
of demand, variation within a day and between days. Figure 7
and through load at each stop in a single profile. The upper,
shows how demand varies between scheduled trips, with
gray step function indicates 85th-percentile segment loads, scheduled trips on the horizontal axis and one measure of
thus adding the dimension of day-to-day variation. This demand, in this case mean boardings, on the vertical. In other
report has already pointed out the importance of extreme versions of this graph (not shown), day-to-day variation is
values of load for both passenger service quality monitoring presented by showing a scatterplot (horizontal whiskers) or
and scheduling. Also shown in Figure 6 for each stop, as well selected percentile values, which allows one to see extreme
as for the route as a whole, are box and whiskers plots of offs values of load that are important to both scheduling and
(just to the left of each stop) and ons. The box extends from operational control. Using established thresholds, trips can be
the 15th percentile to the 85th percentile; a bar indicates the categorized and counted by degree of crowding.
mean value; and the X above the box indicates the maxi- In Figure 7, four of the scheduled trips in the period ana-
mum observed value. lyzed had no valid APC data. They are represented with a
Analysis of demand along a route is necessary for under- large X and a more darkly colored bar whose height is set
standing where along the route high loads occur. It supports equal to the average of the nearest trip before and after it with
decisions about stop relocation and installing stop amenities, valid counts. The issue of imputing values to missing data is
and routing and scheduling actions that affect some parts of discussed in Chapter 11.
a route differently from others, such as short turning, zonal Passenger-miles is another summary measure over a route,
service, and limited stop service (33). being the product of the segment load multiplied by the
Balanced passenger counts (min, mean, 85% and max)
Company: HTM Departure times Dates: 2001/11/05 until 2001/11/08 Trips scheduled: 60 (Count)
Line: 25 From: Lozerlaan From: 07:00 Mon Tue Wed Thu Fri Sat Sun Total Trips used: 35 (58%)
Route: 1 To: Station CS uitstap Until: 09:00 1 1 1 1 0 0 0 4 Trips excluded: 5 ( 8%)
0.3 0.6 2.2 4.1 4.7 6.1 5.6 6.8 7.6 4.1 2.9 6.7
%pmile
0.4 0.8 2.0 5.4 4.1 6.2 4.5 8.9 5.2 7.0 4.0
100
95
90
85
Tritapt 1.0 (b82) license holder is Peter Knoppers, Technische Universiteit Delft. Copyright © 1997-2006 TU Delft
80
75
70
65
60
passengers
55
50
45
40
35
30
25
20
15
10
5
0
EW LL VL DW GL ZP DL HP WZ GM AV SC
LL MS BL He LW AP SP SS VP Br SV NH Total
stop
Source: Delft University, generated by TriTAPT
Figure 6. Load and on/off profile.
OCR for page 38
38
Balanced unlinked passenger trips per vehicle trip (estimated mean 67.9)
Company: HTM Departure times Dates: 2001/11/05 until 2001/11/08 Trips scheduled: 244 (Count)
Line: 25 From: Lozerlaan From: 00:00 Mon Tue Wed Thu Fri Sat Sun Total Trips used: 150 (61%)
Route: 1 To: Station CS uitstap Until: 15:00 1 1 1 1 0 0 0 4 Trips excluded: 19 ( 8%)
Count
2 2 1 1 2 4 4 3 2 3 4 3 0 0 3 0 2 2 0 2 3
3 2 1 4 3 3 3 1 4 3 3 3 4 2 3 2 2 2 4 2
150
145
140
135
130
Tritapt 1.0 (b82) license holder is Peter Knoppers, Technische Universiteit Delft. Copyright © 1997-2006 TU Delft
125
120
115
110
105
100
95
passengers
90
85
80
75
70
65
60
55
50
45
40
35
30
25
20
15
10
5
0
09:00 11:00 13:00
08:00 10:00 12:00 14:00
time [hh:mm]
Source: Delft University, generated by TriTAPT
Figure 7. Boardings by trip across the day.
segment length. When divided by overall route length, this In the future, there may be scheduling tools that account
total indicates the average vehicle occupancy along the route. for within-day and between-day variation in demand, as well
Special considerations relative to measuring passenger-miles as within-day and between-day variation in running time, in
are covered in Chapters 8 and 9. order to design route schedules that respond to how both
An analysis of demand variation across the day supports demand and running times vary across the day, using statis-
scheduling, which, in part, sets headways and departure times tical methods to limit the probability of overcrowding and
so as to achieve target loads. There remains the opportunity insufficient recovery time.
to develop design tools for scheduling that take advantage of
large APC sample sizes to estimate a demand profile across
4.7.3 Passenger Crowding
the day. Using passenger counts combined with measured
headways, and averaging over many days, one should be able There is a strong relationship between vehicle crowding
to derive the passenger arrival rate as a function of time. and passengers' experience of crowding, but the perspectives
Combining these arrival rates in small (e.g., 1-min) time slices are different. For example, if half the trips are empty and half
using a reference frame that moves at the speed of a bus allows are overcrowded, then only 50% of the trips are overcrowded,
one to predict the peak load on a trip based on its departure yet 100% of the passengers experience an overcrowded trip.
time and the departure time of its leader. Measures of crowding from the passenger perspective are dis-
With a minute-by-minute load profile across the day, one cussed in Chapter 7.
valuable tool would be able to find periods of homogeneous
demand within which a constant headway can be used, anal-
4.7.4 Pass-Ups and Special Uses
ogous to scheduling tools that seek periods of homogeneous
running times. Another valuable tool would not assume con- Operator-initiated incident codes used to register such events
stant headways at all, but would select departure times that as pass-ups, wheelchair customers, and bicycle customers can
balance loads between trips, accounting for how demand be used to analyze special demands and events along a route
rates vary across the day, as suggested by Ceder (34). Tools or across the day. Being able to locate them along a route
of this sort are currently under development for the transit might be useful for load analysis, running time analysis, and
agency of the Hague. facility planning.
