Fresno & 3-County Activity-Based Model Training Workshop AB Training... · 2017. 3. 22. · larger daily activity patterns Starting and ending time of activities are modeled choices

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


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


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








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








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



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



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



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



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


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


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


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


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

Documents

Fresno & 3-County Activity-Based Model Training Workshop AB Training... · 2017. 3. 22. · larger daily activity patterns Starting and ending time of activities are modeled choices