Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Fresno & 3-County Activity-Based Model Training Workshop
Joe Castiglione
February 24, 2012
MCAG
Merced, CA
2
Agenda
DaySim-Cube Model System
Overview of activity-based modeling and the “Day Pattern” approach
DaySim-Cube Data Needs
DaySim-Cube Data Preparation
Lunch
DaySim-Cube Operation / Application
Policy Analysis Examples
3
DaySim-Cube Model System
4
Terminology
Activity-based model A travel demand model that produces tours with activity stops
Tours A chain of trips that begin and end at home or work
Trip-based model A travel demand model that produces trips
Advanced models Applied at a disaggregate level, typically with greater spatial
and temporal detail
5
Why use an activity-based model?
Activity-based models…
provide sensitivities to policies and more intuitive
analysis than existing methods
Is more appropriately sensitive to cost, time,
demographics, and policies
produce many performance measures that are not
possible with existing methods
do not necessarily take longer to apply than existing
trip-based methods
Allows for greater spatial and temporal detail
Allows greater household/person attribute detail
6
Differences between trip-based and activity-based models
Contrasting Modeling Approaches Trip-Based
Trips are generated from zonal aggregations of households
Each trip is independent of every other trip’s generation, distribution, mode and timing
Timing/direction of trips is not an explicit choice (fixed factors)
Travel demand is not affected by accessibility or the built environment
Market stratification limited by ability to maintain trip tables throughout model stream
Activity-Based
Simulation of individual households and persons
Trips are chained—modeled as part of tours, sub-tours and larger daily activity patterns
Starting and ending time of activities are modeled choices
Built environment and accessibility variables affect travel demand
Market stratification is a function of individual and household attributes
7
What is different between the SJV trip and AB models?
Much in common Cube framework, tools and user interface
Network methods and assumptions (assignment, skims)
Socioeconomic assumptions (but with more detail in AB)
Primary difference Trip generation, distribution, mode choice replaced with…
Day activity pattern generation (tours and trips)
Destination choice (tours and trips)
Mode choice (tours and trips)
Time-of-day (tours and trips)
8
DaySim-Cube Model System
9
Input Processing Component
10
Skims and Demand Component
11
DaySim Component
12
Assignment Component
13
Activity-Based Modeling and the Day Pattern Approach
14
Activity-Based & Tour-Based Models
Example 2 tours (primary work & work-
based)
5 trips
Day pattern models
Person level
HH level
Condition all subsequent choices of tour and trip/stop destination, mode, and time of day
No NHB!
HOME
WORK
SHOP
MEAL
1
2 3
4
5
Tour and Trip Structure
15
Home-Based Work Trip
Non-Home-Based
Trip
Home-Based
Other Trip
Non-Home-Based Trip
Non-Home-Based Trip
Home-Based Work
(HBW)
Home-Based Other
(HBO)
Non-Home-Based
(NHB)
Zone Prod. Attract. Prod. Attract. Prod. Attract.
1 1 1
2 1 1
3 1 2 1
4 1 1
Total 1 1 1 1 3 3
Zone 1 Zone 3
Zone 2
Zone 4
A Day’s Travel in the 4-step World
16
Home-Based Work Trip
Non-Home-Based
Trip
Home-Based
Other Trip
Non-Home-Based Trip
Non-Home-Based Trip
Zone 1 Zone 3
Zone 2
Zone 4
Work Tour
Primary
Destination
Intermediate
Stop
Origin
Work-Based Tour
Origin Primary
Destination
HH # Per # Tour # Purp Origin
TAZ
Destin.
TAZ
Outbound
Stop1 TAZ
Return
Stop1 TAZ
Mode Sub-
tour
Sub-Tour
Destin.
1023 1 1 Work 1 3 0 2 Transit Yes 4
Data View:
A Day’s Travel in the Activity-Based World
17
DaySim Activity-Based Model System
Detailed travel demand forecasting microsimulation
“Typical” weekday
Regional resident travel
Implemented in multiple regions
Extensively tested and peer reviewed
Features
Simulates 24-hour itineraries
Flexible spatial resolution
30 Minute temporal resolution distributed to minute-by-minute
Tour-based / trip-chaining
Captures effects of time and cost on all travel choices
4-step model components have analogs in activity-based models
INPUT DATA FILES
LONG-TERM CHOICE (once per household)
SHORT-TERM CHOICE
(once per person-day)
OUTPUT FILES
Usual Locations (once per person)
WORK
(Non-Student Workers)
SCHOOL
(All Students)
WORK
(Student Workers)
AUTO OWNERSHIP
(Household)
DAY PATTERN
(activities & home-
based tours for each
person-day)
PRIMARY ACTIVITY
DESTINATIONMAIN MODE PRIMARY ACTIVITY
SCHEDULING
NUMBER & PURPOSE OF
INTERMEDIATE STOPS
ACTIVITY
LOCATIONTRIP MODE
ACTIVITY/TRIP
SCHEDULING
TOURS
(once per
person-tour)
Aggr.Logsums
Aggr.Logsums
Logsums
HALF-TOURS
(twice per person-tour)
INTERMEDIATE STOPS & TRIPS
(once per trip)
RepresentativePopulation
Parcel/PointData
External Trips by Purpose
LOS Skim Matrices, by Periodand Mode (from prior iteration)
TRIP FILE(one record per
person-trip)
TOUR FILE(one record per
person-tour
PERSON FILE(one record per
person-day
18
Activity Purposes in DaySim
Work
School/College
Personal Business (e.g., Medical)
Shopping
Meals
Social/Recreational
Escort Passenger(s)
Home (any activity which takes place within the
home)
19
DaySim Incorporates Greater Detail
Population-based microsimulation procedures loop on
individual households, persons, tours and trips
They do NOT loop on combinations of zones/zone pairs,
population segments, trip purposes, time of day periods
Run time depends mainly on the number of households, & is
not very sensitive to the number of zones, time periods or
population segments distinguished in the simulation.
But, there may be costs in terms of generating and using
related network and land use data
20 20
Greater Spatial Detail in AB Models
Most TAZ systems are too sparse to adequately model:
Effects of localized traffic congestion
Use of walk and bicycle modes and walk access to transit
Effects of changes in urban design and land use
AB can be implemented using any spatial resolution
Fresno, Sacramento: parcels
Northern SJV, San Diego: “blocks”
SF Bay Area, Los Angeles: TAZs
Challenges
Methods to forecast parcel-level land use data
Traffic assignment / skims are typically still coarse (TAZ-level) and network runtimes are typically the performance “bottleneck” in model system
TAZ and block resolution
21
0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
6.0%
7.0%
8.0%
9.0%
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26
% o
f R
egi
on
al T
rave
l
4 PERIOD SKIMS
22 PERIOD SKIMS
EV PMAM MD
1 evening skim
9 hourly midday & shoulder skims
12 30-min peak period skims
Greater Temporal Detail in AB Models
Explicitly represent individual travel across entire day
Interconnected series of tours and trips
Incorporate detail on available “time windows” when scheduling each activity
Network performance can vary within short periods
Resolution
Scheduling models typically use half-hour or hour
Network temporal resolution varies widely
22
AB Scheduling Models
0.0%
2.0%
4.0%
6.0%
8.0%
10.0%
12.0%
14.0%
Be
f…
3:3
0
4:3
0
5:3
0
6:3
0
7:3
0
8:3
0
9:3
0
10
:30
11
:30
12
:30
13
:30
14
:30
15
:30
16
:30
17
:30
18
:30
19
:30
20
:30
21
:30
22
:30
23
:30
0:3
0
1:3
0
2:3
0
Aft…
Axi
s Ti
tle
Work Departure Times
NHTS
DaySim
0.0%
2.0%
4.0%
6.0%
8.0%
10.0%
12.0%
14.0%
16.0%
Be
f…
3:3
0
4:3
0
5:3
0
6:3
0
7:3
0
8:3
0
9:3
0
10
:30
11
:30
12
:30
13
:30
14
:30
15
:30
16
:30
17
:30
18
:30
19
:30
20
:30
21
:30
22
:30
23
:30
0:3
0
1:3
0
2:3
0
Aft…
Work Arrival Times
NHTS
DaySim
0.0%
5.0%
10.0%
15.0%
20.0%
0:0
0
1:0
0
2:0
0
3:0
0
4:0
0
5:0
0
6:0
0
7:0
0
8:0
0
9:0
0
10
:00
11
:00
12
:00
Work Durations
NHTS
DaySim
Scheduling models predict
Desired arrival time / departure time for primary destinations
Arrival /departures times for stops
Key parameters
Person type
Income
Overall day pattern
Available time windows
Network impedances/costs
23
Greater Socio-Demographic Detail in AB Models
Research shows wide variation in behavior related to:
Income
household composition
employment status
Age
other household and person characteristics
Ignoring variation leads to aggregation error and bias
All ABM implementations use synthetic, representative populations
Sample from PUMS / ACS records; control to Census data and available forecasts (HH size, HH income, HH workers).
New methods are evolving; integration with land use models
24
Synthetic Population: Control Data
2 segments Permanent residents
Non-institutional Group quarters population
HH controls Age of head of HH
HH size
HH workers
HH income
Presence of children
Person controls Gender
Age
Data Sources Fresno TAZ data
Northern SJV TAZ data
Census SF1
Census PUMS
ACS PUMS
25
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
$- $5 $10 $15 $20 $25 $30
Value of T ime ($/Hour)
Pro
ba
bil
ity
De
ns
ity
Income $0-30kIncome $30-60k
Income $60-100kIncome $100k+
Socio-Demographic Detail: VOT Distribution
26
DaySim-Cube Data Needs
27
DaySim: Input Files
INPUT DATA FILES
RepresentativePopulation
Parcel/PointData
External Trips by Purpose
LOS Skim Matrices, by Periodand Mode (from prior iteration)
Representative or “synthetic” population of the region’s residents
Detailed parcel and point data provides more accurate representation of real travel times and cost
External or auxiliary trips are not predicted by DaySim, but are a critical part of the region’s travel market
Externals
Trucks
Airports
LOS (level-of-service) skim matrices capture travel times, costs and other relevant attributes by travel submodel and time of day
28
Parcel / Microzone Data
DaySim uses parcels or microzones as a fundamental spatial units
Attributes include:
Location
Area
Housing units
Enrollment by school type
Employment by sector
Transportation network access
Urban form measures
Offstreet parking
Buffers of housing units, enrollment, employment
Fresno Total Employment by Parcel
29
Parcel / Block Data: Buffers
Buffers prepared as urban form indicators ½ mile
¼ mile
Distance decay functions
Use parcel centroids
Attributes buffered Housing units
Employment by sector
Enrollment
Street intersections by type (deadend, 3-way, 4-way)
30
Fresno Parcel Buffer: Access to Retail Employment
31
Northern SJV Block Buffer: Access to Retail
Detailed housing and employment information limited to TAZ resolution
Want sensitivity to small-scale/local conditions
Implementing models using Census block-based geography
TAZ totals disaggregated to blocks using
Decennial Census
LED/LEHD
32
Parcel / Block Data: Transit Access
TAZ and parcel/block-based information used to estimate network impedances
Transit access is refined using parcel-level access to transit by submode
Other enhancements to refine TAZ-based nonmotorized impedances
33
Fresno Parcel Data: Transit Access
34
Parcel/Block Data: Street Connectivity
Measures of intersection or nodes by type within ½ and ¼ mile buffers
Types Deadends
T-intersections
Traditional intersections
Based on all-streets network
35
Fresno Parcel Buffer: Street Connectivity
36
Northern SJV Block Buffer: Grade School Enrollment
37
DaySim-Cube Data Preparation
38
Synthetic Population: PopGen
Open source
Supports use of person-level and HH-level controls
Easy-to-use
Flexible
GUI
Output visualization
Steps
Prepare control data
Prepare sample data
Synthesize population
Allocate to parcels/blocks
39
Datafile Preparation Tools: Synthetic Population
List of regional resident households and persons
Based on observed or forecating distributions of socioeconomic variables
Created by sapling detailed Census microdata
Basis for all subsequent trip-making in themodel
Segments
Permanent HHs and persons
Group quarters residents
Variable Definition
HHNO Household id
HHSIZE Household size
HHVEHS Vehicles available
HHWKRS Household workers
HHFTW HH full time workers (type 1)
HHPTW HH part time workers (type 2)
HHRET HH retired adults (type 3)
HHOAD HH other adults (type 4)
HHUNI HH college students (type 5)
HHHSC HH high school students (type 6)
HH515 HH kids age 5-15 (type 7)
HHCU5 HH kids age 0-4 (type 8)
HHINCOME Household income ($)
HOWNRENT Household own or rent
HRESTYPE Household residence type
HHPARCEL Residence parcel id
HHEXPFAC HH expansion factor
SAMPTYPE Sample type
HH file format
40
Datafile Preparation Tools: Synthetic Population
List of regional resident households and persons
Based on observed or forecating distributions of socioeconomic variables
Created by sapling detailed Census microdata
Basis for all subsequent trip-making in themodel
Segments
Permanent HHs and persons
Group quarters residents
Variable Definition
HHNO hh id
PNO person seq no on file
PPTYP person type
PAGEY age in years
PGEND gender
PWTYP worker type
PWPCL usual work parcel id
PSTYP student type
PSPCL usual school parcel id
PUWMODE usual mode to work
PUWARRP Usual arrival period to work
PUWDEPP Usual depart period from work
PTPASS transit pass?
PPAIDPRK paid parking at workplace?
PDIARY Person used paper diary?
PPROXY proxy response?
PSEXPFAC Person expansion factor
Person file format
41
Synthetic Population Validation
Table 1. Observed Permanent Households by Size
1 2 3 4+ Total
Fresno 62,120 77,661 45,404 100,192 285,377
Merced 14,852 22,032 13,530 31,461 81,875
San Joaquin 43,034 58,046 34,035 72,365 207,480
Stanislaus 34,614 49,747 27,952 56,762 169,075
Total 154,620 207,486 120,921 260,780 743,807
Table 2. Estimated Permanent Households by Size
1 2 3 4+ Total
Fresno 62,195 77,596 45,540 100,046 285,377
Merced 14,876 22,002 13,570 31,427 81,875
San Joaquin 43,040 57,979 34,194 72,267 207,480
Stanislaus 34,597 49,679 28,183 56,616 169,075
Total 154,708 207,256 121,487 260,356 743,807
Table 3. Difference in Permanent Households by Size
1 2 3 4+ Total
Fresno 75 -65 136 -146 0
Merced 24 -30 40 -34 0
San Joaquin 6 -67 159 -98 0
Stanislaus -17 -68 231 -146 0
Total 88 -230 566 -424 0
Table 4. Observed Permanent Household Population by Age
0-4 5-17 18-24 25-54 55-64 65+ Total
Fresno 76,657 187,010 90,456 336,099 74,823 81,436 846,481
Merced 24,756 59,229 27,739 99,625 20,589 24,621 256,559
San Joaquin 51,561 131,209 63,384 246,920 60,295 69,499 622,868
Stanislaus 44,444 104,915 48,408 200,352 47,792 55,550 501,461
Total 197,418 482,363 229,987 882,996 203,499 231,106 2,227,369
Table 5. Estimated Permanent Household Population by Age
0-4 5-17 18-24 25-54 55-64 65+ Total
Fresno 76,613 185,199 90,523 342,876 76,637 82,727 854,575
Merced 23,898 58,678 27,066 96,380 19,873 23,883 249,778
San Joaquin 50,810 130,013 62,228 241,267 58,650 67,754 610,722
Stanislaus 41,266 103,265 47,598 196,645 46,522 54,445 489,741
Total 192,587 477,155 227,415 877,168 201,682 228,809 2,204,816
Table 6. Difference in Permanent Household Population by Age
0-4 5-17 18-24 25-54 55-64 65+ Total
Fresno -44 -1,811 67 6,777 1,814 1,291 8,094
Merced -858 -551 -673 -3,245 -716 -738 -6,781
San Joaquin -751 -1,196 -1,156 -5,653 -1,645 -1,745 -12,146
Stanislaus -3,178 -1,650 -810 -3,707 -1,270 -1,105 -11,720
Total -4,831 -5,208 -2,572 -5,828 -1,817 -2,297 -22,553
Household-level Validation Person-level Validation
42
Datafile Preparation Tools: 3-County Microzones
Basic Spatial Unit for referencing socioeconomic data such as HHs, population, employment
Essentially Census block-level geography
Requires
TAZ-level controls of key employment and socioeconomic attributes prepared by agency
Block-level household and employment data derived from free, publically available data
2 tools developed to automate the preparation of 3-County DaySim inputs
MicrozoneDistribution: creates microzone totals of households, population, and employment
ParcelBuffer: Prepare derived measures at microzone level such as buffers and transit access
43
Datafile Prep Tools: Microzone Distribution Inputs
Microzone distribution tool inputs
TAZ file
Block file
TAZ-Block intersect file
School file
44
Microzone Distribution Inputs: TAZ file
FIELD DESCRIPTION
TAZ taz number
XCOORD X coordinate of taz centroid – state plane feet
YCOORD Y coordinate of taz centroid – state plane feet
SQFT taz area – square feet
HH households in taz
STUGRD grade school enrollment in taz
STUHGH high school enrollment in taz
STUUNI university enrollment in taz
EMPEDU education employment in taz
EMPFOOD food employment in taz
EMPGOV government employment in taz
EMPIND industrial employment in taz
EMPMED medical employment in taz
EMPOFC office employment in taz
EMPRET retail employment in taz
EMPSVC service employment in taz
EMPOTH other employment in taz
EMPTOT total employment in taz
Base on inputs to the trip-based model
Employment reclassified from 21 detailed MIP sectors to 9 core sectors
45
Microzone Distribution Inputs: Block file
Contains key information used to disaggregate TAZ info to microzone level
Census / ACS
HH info
LED database
Employment info
2-digit NAICS
Can be adjusted to reflect different assumptions
FIELD DESCRIPTION
ID Block id number
XCOORD X coordinate of block centroid – state plane feet
YCOORD Y coordinate of block centroid – state plane feet
SQFT block area – square feet
HH households in block
STUGRD grade school enrollment in block
STUHGH high school enrollment in block
STUUNI university enrollment in block
EMPEDU education employment in block
EMPFOOD food employment in block
EMPGOV government employment in block
EMPIND industrial employment in block
EMPMED medical employment in block
EMPOFC office employment in block
EMPRET retail employment in block
EMPSVC service employment in block
EMPOTH other employment in block
EMPTOT total employment in block
46
Microzone Distribution Inputs: TAZ-block intersect file
Source for microzone geography
User specified minimum size
FIELD DESCRIPTION
ID Intersect id number
XCOORD X coordinate of intersect centroid – state plane feet
YCOORD Y coordinate of intersect centroid – state plane feet
AREA intersect area – square feet
TAZID TAZ in which intersect is located
BLOCKID Block in which intersect is located
47
Microzone Distribution Inputs: School file
Detailed info on school locations and enrollment available
No need for block-level controls for disaggregating
FIELD DESCRIPTION
TAZ taz number
XCOORD X coordinate of taz centroid – state plane feet
YCOORD Y coordinate of taz centroid – state plane feet
SQFT taz area – square feet
HH households in taz
STUGRD grade school enrollment in taz
STUHGH high school enrollment in taz
STUUNI university enrollment in taz
48
Microzone Distribution Outputs
Contains all fields required for input to parcel/microzone buffer tool
FIELD DESCRIPTION
microzoneid Microzone ID number
xcoord_p X coordinate – state plane feet
ycoord_p Y coordinate – state plane feet
sqft_p microzone area – square feet
taz_p corresponding TAZ number
block_p corresponding census block number
hh_p households on microzone
stugrd_p grade school enrollment on microzone
stuhgh_p high school enrollment on microzone
stuuni_p university enrollment on microzone
empedu_p educational employment on microzone
empfoo_p food employment on microzone
empgov_p government employment on microzone
empind_p industrial employment on microzone
empmed_p medical employment on microzone
empofc_p office employment on microzone
empret_p retail employment on microzone
empsvc_p service employment on microzone
empoth_p other employment on microzone
emptot_p total employment on microzone
parkdy_p offstreet daily parking on microzone
parkhr_p offstreet hourly parking on microzone
ppricdyp offstreet daily parking price
pprichrp offstreet hourly parking price
49
Microzone Distribution: Using the tool
User configurable control file
Application called from console (DOS prompt)
Runs in seconds
50
Datafile Prep Tools: Parcel / Microzone Buffer Inputs
Parcel / Microzone buffer tool inputs
Parcel / microzone base file
Intersection file
Transit stop file
Openspace file
51
Parcel / Microzone Buffer Inputs: Base file
FIELD DESCRIPTION
TAZ taz number
XCOORD X coordinate of taz centroid – state plane feet
YCOORD Y coordinate of taz centroid – state plane feet
SQFT taz area – square feet
HH households in taz
STUGRD grade school enrollment in taz
STUHGH high school enrollment in taz
STUUNI university enrollment in taz
EMPEDU education employment in taz
EMPFOOD food employment in taz
EMPGOV government employment in taz
EMPIND industrial employment in taz
EMPMED medical employment in taz
EMPOFC office employment in taz
EMPRET retail employment in taz
EMPSVC service employment in taz
EMPOTH other employment in taz
EMPTOT total employment in taz
Parcel or microzone-level
Based on parcel file or microzone output file
Core attributes
HHs
Employment
Enrollment
QA/QC key
52
Parcel / Microzone Buffer Inputs: Intersection file
Used to calculate urban form measure: types of intersections within different buffers
Deadends
T-intersections
Traditional intersections
“All streets” network-based
GIS-based tool to develop intersection by type data
Alternative methods for developing future year all-streets assumptions
FIELD DESCRIPTION
id Intersection ID number
links Number of links associated with node
xcoord_p X coordinate – state plane feet
ycoord_p Y coordinate – state plane feet
53
Parcel / Microzone Buffer Inputs: Transit stop file
Used to calculate distance to transit
Refinements to skims
Urban form measure
Based on shapefiles of transit stop locations and demand-responsive route alignments
Future year locations must be assumed if no info available
FIELD DESCRIPTION
id Transit stop ID number
mode Transit submode code
xcoord_p X coordinate – state plane feet
ycoord_p Y coordinate – state plane feet
54
Parcel / Microzone Buffer Inputs: Openspace file
Publically accessible open space
Based on CPAD: California Protected Areas Database
Typically not included in travel models due to complications associated with use as a size variable
FIELD DESCRIPTION
id Open space grid ID number
xcoord_p X coordinate – state plane feet
ycoord_p Y coordinate – state plane feet
sqft Open space grid cell size in sq ft
55
Parcel / Microzone Buffer Outputs
Contains all fields required for input to AB model
FIELD DESCRIPTION
id Microzone/parcel ID number
xcoord_p X coordinate – state plane feet
ycoord_p Y coordinate – state plane feet
sqft_p Area – square feet
taz_p TAZ number
lutype_p land use type
hh_p, 1, 2 households on microzone/parcel, buffer 1, buffer 2
stugrd_p 1, 2 grade school enrollment on microzone/parcel buffer 1, buffer 2
stuhgh_p 1, 2 high school enrollment on microzone/parcel buffer 1, buffer 2
stuuni_p 1, 2 university enrollment on microzone/parcel buffer 1, buffer 2
empedu_p 1, 2 educational employment on microzone/parcel buffer 1, buffer 2
empfoo_p 1, 2 food employment on microzone/parcel buffer 1, buffer 2
empgov_p 1, 2 government employment on microzone/parcel buffer 1, buffer 2
empind_p 1, 2 industrial employment on microzone/parcel buffer 1, buffer 2
empmed_p 1, 2 medical employment on microzone/parcel buffer 1, buffer 2
empofc_p 1, 2 office employment on microzone/parcel buffer 1, buffer 2
empret_p 1, 2 retail employment on microzone/parcel buffer 1, buffer 2
empsvc_p 1, 2 service employment on microzone/parcel buffer 1, buffer 2
empoth_p 1, 2 other employment on microzone/parcel buffer 1, buffer 2
emptot_p 1, 2 total employment on microzone/parcel buffer 1, buffer 2
parkdy_p 1, 2 offstreet daily parking on microzone/parcel buffer 1, buffer 2
parkhr_p 1, 2 offstreet hourly parking on microzone/parcel buffer 1, buffer 2
ppricdyp 1, 2 offstreet daily parking price microzone/parcel buffer 1, buffer 2
pprichrp 1, 2 offstreet hourly parking price microzone/parcel buffer 1, buffer 2
56
Parcel / Microzone Buffer Outputs
Contains all fields required for input to AB model
nodes1_1, 2 number of single link street nodes (dead ends) within buffer 1, buffer 2
nodes3_1, 2 number of three-link street nodes (T-intersections) within buffer 1, buffer 2
nodes4_1, 2 number of 4+ link street nodes (traditional 4-way +) within buffer 1, buffer 2
tstops_1, 2 number of transit stops within buffer 1, buffer 2
nparks_1, 2 number of open space parks within buffer 1, buffer 2
aparks_1, 2 open space area in swuare feet within buffer 1, buffer 2
dist_lbus distance to nearest local bus stop from microzone/parcel
dist_ebus distance to nearest express bus stop from microzone/parcel
dist_crt distance to nearest commuter rail stop from microzone/parcel
dist_fry distance to nearest ferry stop from microzone/parcel
dist_lrt distance to nearest light rail stop from microzone/parcel
dist_park distance to nearest park from microzone/parcel
57
Parcel / Microzone Buffer : Using the tool
User configurable control file
Application called from console (DOS prompt)
Runs in minutes
58
DaySim-Cube Operation / Application
59
DaySim-Cube Model System
60
Run DaySim-Cube Application
61
Run Component
62
Scenario Editor
63
Scenario Editor
64
Scenario Editor
65
Scenario Editor
66
Scenario Editor
67
Scenario Editor
68
Input Processing Component
69
Skims and Demand Component
70
DaySim Component
71
Assignment Component
72
DaySim: Outputs
In the same general form as
household travel diary data with
the following files
Household
Person
Personday
Trip
Tour
Outputs converted to matrices
and used with Cube and other
traditional equilibrium
assignment tools using any time
period definition
SAMPN PERSN TOURNO TOURHALF TRIPNO OTAZ OCEL DTAZ DCEL OPURP DPURP DEPTIME ARRTIME EACTTIM TRAVTIM TRAVDIST EXPFACT
1 1 1 1 1 445 429711 1088 133524 8 4 1222 1238 1556 16.09 8.56 1.00
1 1 1 2 1 1088 133524 445 429711 4 8 1556 1615 2659 18.65 8.56 1.00
DaySim Trip List Output Example
Trip-based Matrix Output Example
73
DaySim: Trip Outputs
Spatially, temporally,
behaviorally detailed output
Can be used in combination
with other inputs/outputs to
provide new analysis
capabilities
Variable Definition
HHNO Household id
PNO person seq no on file
DAY Diary / simulation day ID
TOUR tour id
HALF tour half
TSEG trip seqgment no within half tour
TSVID original survey trip id no.
OPURP trip origin purpose
DPURP trip dest purpose
OADTYP trip origin address type
DADTYP trip destination address type
OPCL trip origin parcel
OTAZ trip origin zone
DPCL trip dests parcel
DTAZ trip dest zone
MODE trip mode
PATHTYPE transit submode
DORP trip driver or passenger
DEPTM trip deparute time (min after 3 am)
ARRTM trip arrival time (min after 3 am)
ENDACTTM*** trip dest activity end time
TRAVTIME network travel time, min (by sov)
TRAVCOST network travel time, min (by sov)
TRAVDIST network travel distance, miles (by sov)
TREXPFAC trip expansion factor
74
Policy Analysis Examples
75
More Performance Measures
Activity-based model raw outputs are disaggregate trip
records, with important identifying attributes:
Activity/trip purpose, start/end times, travel mode, location
IDs
Tour purpose, primary location, primary mode, start/end times
Household ID, Person ID, Tour ID, Trip/Activity ID
This allows the user to summarize system performance
data along a at least four potentially useful dimensions:
Household and person attributes
Time period of the day
Activity/trip/tour purposes
Geographic units and spatial clusters
76
Ability to Derive Performance Measures
Shopping Trip Frequency
Time Period
District
Work Activity Arrival/Depar
ture Times District
Mean Trip Length
Age Group Time
Period
Trips Per Tour
Gender Value of
Time
Mode Share Income Group
Trip Purpose
Mode Share of Persons
Within ¼-mile of Transit
Parcels Walk
Trips/Person
Tolls paid Trip
Purpose TAZ
Can summarize travel
behavior metrics by
various combinations
of the activity-based
model dimensions
Some examples are
77
Environment and Climate Change Sensitivities
Disaggregate data on travel provides more accurate
estimates of emissions
Trip chaining provides better data on starts/stops
Compact Urban Form and Transit Oriented
Development represented more completely through
greater level of detail
Pricing and TDM are important policies for GHG
reduction
Vehicle ownership (type, age) affects emissions
77
78 78
GHG estimates by residence parcel -- Sacramento Area Council of Governments
Example: Environment and Climate Change
79
Policies: Transit Destination and mode choices for round trips (tours)
affect destination and mode choices for individual
trips
Tour-level destination and mode choices consider both
outbound and return availability, travel times and
costs
Added detail from home to the transit stop and from
the stop to the destination and for local walk and bike
travel has improved accuracy
Transit fare passes and driver’s licenses can be
explicitly represented
Built environments affect station area ridership
79
Transit Policy Sensitivities
80
Example: Transit New Starts San Francisco
Work Tour Destination-Based User Benefit
•San Francisco Central
Subway
•1.4 miles connecting
South of Market to
Chinatown
• Third Street LRT 7.1
mile surface line (IOS =
Baseline)
81
Example: Transit New Starts Sacramento
Sacramento State BRT
Activity-based model used to
simulate campus arrivals and
departures by ½ hour time periods
Parking lots fill up -> park further
from destination
Choice of BRT or walk from lot to
destination
82
BRT Boardings By Time Period
0
100
200
300
400
500
600
5:0
0
6:3
0
8:0
0
9:3
0
11
:00
12
:30
14
:00
15
:30
17
:00
18
:30
20
:00
21
:30
23
:00
Time Period
Bo
ard
ing
s
BRT Boardings
Total Available Parking By Time Period
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
5:00
6:30
8:00
9:30
11:00
12:30
14:00
15:30
17:00
18:30
20:00
21:30
23:00
Total Spaces
AB model tracks
time in ½ hour
periods
Parking constraints
and policies affect
transit ridership
Example: Sacramento BRT PNR by Time of Day
83
Pricing Sensitivities
Ability to represent time-cost tradeoffs on multiple,
relevant travel choices:
Daily/trip choices: route, time of day, mode, location,
vehicle occupancy, pay toll/avoid toll, parking
Long-term choices: work and school location, vehicle
ownership, transit pass holding
Affected by income, household structure and mobility
resources
83
84 84
Central
Business
District
Congestion Pricing
Zone Boundary
Type of Driver/ Group Level of Discount
Taxi, Transit FREE
Commercial Vehicles, Shuttles
FLEET
Rental Cars & Car Sharing FLEET
Toll-payer ‘Fee’-bate $1 off
Low-Income (Lifeline Value) 50% off
Disabled Drivers 50% off
Zone Residents 50% off
Low-Emission Vehicles -
HOV/Carpool -
May be accompanied by
investment in Means-Based
Fare Assistance Program
Helps minimize administrative
impacts for businesses, and
keeps industry moving
Would require
documentation of
inability to take transit
Example: Congestion Pricing
85
Travel Demand Management Sensitivities
Strategies to change travel behavior in order to reduce congestion and improve mobility
Telecommuting\Work-at-home
Flexible work schedules (off-peak)
Rideshare programs
Scenario-based approaches necessary
Model system captures the effects of TDM policy outcomes
Cannot identify which policies will affect flexible work schedules
But can estimate the impact on transportation system performance of shift from a 5-day 8-hour work week to a 4-day 9+ hour work week
86
Example: Travel Demand Management
• “Flexible Schedule” scenario
• Asserted assumptions about:
• Fewer individual work activities
• Longer individual work durations
• Aggregate work durations constant
• Target: Fulltime Workers
0
1
2
3
4
5
6
7
8
Du
rati
on
1.0
0
2.0
0
3.0
0
4.0
0
5.0
0
6.0
0
7.0
0
8.0
0
9.0
0
10
.00
11
.00
12
.00
13
.00
14
.00
15
.00
% o
f To
urs
Work Tour Duration Distribution
Original
Adjusted
Tours by Purpose (Fulltime Workers)
Original Adjusted Adj/Orig
Work 94,408 78,472 0.83
School 115 140 1.22
Escort 8,070 9,023 1.12
Pers Bus 13,519 16,848 1.25
Shop 10,531 12,938 1.23
Meal 3,817 3,842 1.01
Soc/Rec 13,076 14,360 1.10
Workbased 27,949 23,211 0.83
Total 171,485 158,834 0.93
87
TDM: Demand Impacts
~4% Reduction in overall trips
Reduced peak period and midday travel
More early AM travel and evening travel
Fewer, and earlier, work trips
More nonwork trips in morning and evening with fewer in midday
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
03
:00
04
:00
05
:00
06
:00
07
:00
08
:00
09
:00
10
:00
11
:00
12
:00
13
:00
14
:00
15
:00
16
:00
17
:00
18
:00
19
:00
20
:00
21
:00
22
:00
23
:00
00
:00
01
:00
02
:00
Difference in Trips by Time of Day
TDM
-4000
-3000
-2000
-1000
0
1000
2000
3000
4000
03
:00
04
:00
05
:00
06
:00
07
:00
08
:00
09
:00
10
:00
11
:00
12
:00
13
:00
14
:00
15
:00
16
:00
17
:00
18
:00
19
:00
20
:00
21
:00
22
:00
23
:00
00
:00
01
:00
02
:00
Difference in Trips by Time of Day
TDM-WORK
TDM-NONWORK
88
TDM: Supply Impacts
Total VMT declines slightly
Reduced peak period and midday VMT, increased VMT in evening
Reduced peak period and midday delay across all facility types, additional delay in the evening
0
50000
100000
150000
200000
250000
300000
0:0
0
1:0
0
2:0
0
3:0
0
4:0
0
5:0
0
6:0
0
7:0
0
8:0
0
9:0
0
10
:00
11
:00
12
:00
13
:00
14
:00
15
:00
16
:00
17
:00
18
:00
19
:00
20
:00
21
:00
22
:00
23
:00
30-minute time period
VMT by 30 Minute Period
BASE
TDM
0
200
400
600
800
1000
0:0
0
1:0
0
2:0
0
3:0
0
4:0
0
5:0
0
6:0
0
7:0
0
8:0
0
9:0
0
10
:00
11
:00
12
:00
13
:00
14
:00
15
:00
16
:00
17
:00
18
:00
19
:00
20
:00
21
:00
22
:00
23
:00
30-minute time period
Hours of Delay - Major Arterials
BASE
TDM
0
50
100
150
200
250
300
0:0
0
1:0
0
2:0
0
3:0
0
4:0
0
5:0
0
6:0
0
7:0
0
8:0
0
9:0
0
10
:00
11
:00
12
:00
13
:00
14
:00
15
:00
16
:00
17
:00
18
:00
19
:00
20
:00
21
:00
22
:00
23
:00
30-minute time period
Hours of Delay - Minor Arterials
BASE
TDM
0
100
200
300
400
500
0:0
0
1:0
0
2:0
0
3:0
0
4:0
0
5:0
0
6:0
0
7:0
0
8:0
0
9:0
0
10
:00
11
:00
12
:00
13
:00
14
:00
15
:00
16
:00
17
:00
18
:00
19
:00
20
:00
21
:00
22
:00
23
:00
30-minute time period
Hours of Delay - Collectors
BASE
TDM
89
Common Concerns About AB Models
The models are more complicated and hard to
understand
Only to “veterans”
For students, ABM are more intuitive
The models are so complicated that they may not
reflect behavior realistically
Model components can be integrated via logsums
The methods are still evolving (a moving target)
Recent applications evolving toward a common structure
The model results contain some random variation
Can be minimized using multiple runs and random number synchronization
across scenarios. Helps avoid false precision in interpreting results.
90
Requires software in addition to standard network packages The main software approaches are open source, and becoming more user-
oriented over time
The models take longer to run Network assignment is typically the performance bottleneck in model
system, esp with more temporal detail
AB model software improvements and hardware advances(multi-processing, more memory) have significantly reduced demand-side runtimes
Common Concerns about AB Models
91
Thanks!
92
Differences between AB & 4-step modeling
Units of decisions
Trips vs. Trips / Tours / Person-days / Household-days
Method of predicting choices
Top-down aggregate shares vs. Bottom-up microsimulation
Amount of detail that can be accommodated
Socio-Demographic: A few segmentations vs. Many variables
Temporal: Broad time periods vs. Hours or half-hours
Spatial: Zones vs Parcels or points
AB models are less familiar to potential users
93
Multiplying rates and fractions
Trip generation Households by population segment / residence TAZ
Trip distribution Trips by pop. segment / trip purpose / O-D TAZ pair
Time-of-day Trips by segment / trip purpose / O-D pair / time period
Mode choice Trips by segment / trip purpose / O-D pair / time period / mode
Many millions of numbers, mostly small fractions of
trips
Top-Down Forecasting (aggregate 4-step)
94
Bottom-up forecasting (population microsimulation)
Adding up discrete choices
Apply a hierarchical series of models to predict
behavior at several different levels for each
representative household and person in the regional
population: Work and school locations
Auto ownership
Household-days
Person-days
Tours & Trips
Origin and destination locations
Departure time and arrival time
Mode used
Millions of trip records, each a single trip >> ADD
THEM UP